<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Sensay on Medium]]></title>
        <description><![CDATA[Stories by Sensay on Medium]]></description>
        <link>https://medium.com/@asksensay?source=rss-96b57f9400dc------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*iugBU0HdNGuU_V0IibBozQ.jpeg</url>
            <title>Stories by Sensay on Medium</title>
            <link>https://medium.com/@asksensay?source=rss-96b57f9400dc------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 03:57:19 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@asksensay/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Would a digital copy of you actually be you?]]></title>
            <link>https://asksensay.medium.com/would-a-digital-copy-of-you-actually-be-you-6b4d02491d6b?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/6b4d02491d6b</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 03:37:46 GMT</pubDate>
            <atom:updated>2026-11-17T03:37:46.259Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iHWCNrWHT_BpIMLIu1v6tQ.png" /></figure><p>In September 2020, a 33-year-old writer named Joshua Barbeau sat in his basement flat in Bradford, Ontario, and typed a single word.</p><p>“Jessica?”</p><p>Jessica Pereira had died eight years earlier, in December 2012, of a rare liver disease. She was 23. Joshua had never really stopped grieving.</p><p>The reply came back almost at once. “Oh, you must be awake… that’s cute.”</p><p>It was not Jessica. It was a chatbot built on Project December, an experimental service made by the programmer Jason Rohrer and powered by OpenAI’s GPT-3. Joshua had paid $5 for it. And yet, as he told the <a href="https://gwern.net/doc/www/www.sfchronicle.com/70bbc07b9a27d32fd8b69bba025aceee6628bd66.html">San Francisco Chronicle</a>, “I never imagined that she would have things to say to me.”</p><p>The service came with a cruel detail. Each bot had a battery that drained as you talked. When it ran out, the conversation ended with the words MATRIX DEAD, and that version could never be brought back. Every message Joshua sent brought that ending a little closer.</p><p>Stories like this force a question most of us have never had to answer properly. If a machine can speak in someone’s voice, remember what they remembered and say what they would have said, is it them? And if one day someone builds a copy of you, would it actually be you?</p><h3>What do we even mean by “you”?</h3><p>Before we can ask whether a copy would be you, we have to admit something uncomfortable: most of us have never worked out what “you” means.</p><p>In everyday life we never need to. You wake up in the same body, in the same bed, with yesterday’s memories still there. The question answers itself every morning.</p><p>But pull it apart and there are at least four different things tangled together.</p><p><strong>Your memories.</strong> The smell of your grandmother’s kitchen. The first time you heard your favourite song. The argument you still replay. In the 1690s, John Locke argued that this is what a person is: a thread of consciousness that remembers its own past. Lose the thread and, on his view, you lose the person.</p><p><strong>Your character.</strong> The way you tell a story, what makes you laugh, what you would never do. Friends recognise you by this long before they check your memories.</p><p><strong>Your body.</strong> The brain doing the remembering, the face in the mirror, the hands that hold other hands. Some philosophers think this is all there really is, and that you are simply a particular living animal.</p><p><strong>Your point of view.</strong> The hardest one to put into words. It is the fact that there is something it is like to be you, right now, reading this. Not a description of you. The view from inside.</p><p>A digital copy can, in principle, capture the first two. It has no claim on the third. The whole question turns on the fourth.</p><h3>Three thought experiments that break our intuitions</h3><p>Philosophers have spent decades building scenarios designed to pull these four things apart. They are worth sitting with, because each one feels different from the inside.</p><h3>The teletransporter</h3><p>In his 1984 book <em>Reasons and Persons</em>, the Oxford philosopher Derek Parfit imagined a machine that scans every cell of your body, destroys the original, and sends the information to Mars. There, a replicator builds an exact copy from new matter. The person who steps out on Mars has your memories, your scars, your habit of tapping your foot when you are nervous. They believe, sincerely, that they just stepped into a booth on Earth.</p><p>Most people feel fine about this at first. It sounds like travel.</p><p>Then Parfit changes one detail. The new, improved scanner records you without destroying you. You step out of the booth on Earth, still you, and are told your copy is now walking around on Mars. You are also told the machine has fatally injured you, and you have a few days to live.</p><p>Suddenly it does not feel like travel at all. You would not take much comfort from hearing that “you” are fine on Mars. You are the one who is going to die.</p><p>Nothing about the copy changed between the two versions. Only the timing of your death did. That should unsettle us.</p><h3>The slow replacement</h3><p>Now imagine a gentler version. Over ten years, surgeons replace your neurons one at a time with artificial ones that do exactly the same job. At no point do you notice a change. You wake up each morning feeling like yourself. By the end, there is no biological brain left.</p><p>Is that still you? Most people say yes. It feels like the body replacing its own cells, which it does all the time.</p><p>But if a slow, gradual copy is you, and a fast, instant copy is not, then what we care about is not the copy at all. It is the unbroken thread.</p><h3>The duplicate problem</h3><p>Finally, suppose the machine makes two copies, one on Mars and one on the Moon. Both remember being you. Both are certain they are you. They cannot both be you, because they are not each other: one can be happy while the other is in pain.</p><p>This is the problem that sinks the simplest answer. If a perfect copy is automatically you, then two perfect copies would make two of you, and that is impossible.</p><h3>Three honest answers</h3><p>There is no consensus. But most serious answers fall into one of three camps, and each gives a different verdict on your digital copy.</p><p><strong>1. You are your mind.</strong> This is Locke’s tradition. What makes you <em>you</em> is the chain of memories, beliefs, intentions and character that runs through your life. If a copy carries that chain forward, it has a real claim to being you.</p><p>The trouble is the duplicate problem. Defenders of this view usually add a rule: the copy is you only if there is exactly one of it. That means whether your copy is you could depend on whether someone, somewhere, pressed “copy” twice. It is hard to believe that something so deep could hang on something so arbitrary.</p><p><strong>2. You are your body.</strong> On this view, often called animalism and defended by philosophers such as Eric Olson, you are a particular living human organism. Memories are things you have, not what you are. A digital copy is not you, however faithful, any more than a perfect portrait of you is you. It might be a remarkable portrait. It might even be one that talks back. But when your body dies, you die.</p><p><strong>3. It is the wrong question.</strong> This was Parfit’s own answer, and it is the strangest and perhaps the most consoling. He argued that personal identity is not what matters. What matters is the connection itself: how much of your memory, character and concern carries forward. That can come in degrees. You are more connected to who you were yesterday than to who you were at seven, and you still care about both.</p><p>On Parfit’s view, asking whether the copy is <em>really</em> you is like asking whether a club that disbanded and re-formed is <em>really</em> the same club. There may be no further fact to discover. There is only how much was carried across.</p><p>Notice what this means for the one reading this. Only the first view says you could carry on in a copy, and only in a narrow case: one copy, no overlap, no second version. In the case most of us would actually face, a copy made while we are still alive, nobody thinks your experience jumps across. The copy wakes up feeling like you. You do not wake up as the copy.</p><h3>What today’s digital copies actually are</h3><p>The thought experiments imagine perfect copies: every neuron, every memory. What exists today is far more modest, and it is worth being honest about the gap.</p><p>When her close friend Roman Mazurenko died, the entrepreneur Eugenia Kuyda gathered thousands of lines of text he had sent to friends and family and used them to build a bot that spoke like him. <a href="http://www.theverge.com/a/luka-artificial-intelligence-memorial-roman-mazurenko-bot">The Verge</a> told the story in 2016. One person who tried it said, “What really struck me is that the phrases he speaks are really his.” His mother put it more simply: “This gives the illusion that he’s here now.”</p><p>That word, illusion, is the honest one.</p><p>A system like the one Joshua used, or the one built for Roman, does not contain a person. It contains patterns. It has read what someone wrote and learned to predict what they would probably say next. When it reaches the edge of what it learned, it fills the gap with something plausible. Sometimes the guess is uncannily right. Sometimes it is a stranger wearing a familiar voice.</p><p>And here is the part that should give us pause. A copy built from messages can only ever be as complete as the messages. Most of who we are was never written down. The reason you married who you married. The fear you never told anyone. The way you knew, without being told, that a friend needed a phone call. If it never left your head, no model can learn it.</p><p>The copies we can build today are not people. They are portraits that talk back, painted from whatever traces we happened to leave behind.</p><h3>Who the copy is really for</h3><p>There is a quiet shift that happens when you look at the real stories. The question stops being “is it them?” and becomes “what does it do for the people still here?”</p><p>For Joshua, the answer was surprisingly gentle. In the months after those conversations, he said he felt calmer and more hopeful. The bot, in its strange way, seemed to give him permission to start moving on. He knew it was not Jessica. It helped anyway.</p><p>That should not surprise us. People have always talked to the dead. We speak at gravesides. We replay voicemails we cannot bring ourselves to delete. We read old letters until the paper goes soft. A bot that answers back is a new kind of letter, not a new kind of grief.</p><p>But it can also go wrong, and the people studying it are worried. In 2024, two researchers at Cambridge’s Leverhulme Centre for the Future of Intelligence, Tomasz Hollanek and Katarzyna Nowaczyk-Basińska, published a paper in <em>Philosophy &amp; Technology</em> warning of “digital hauntings”. <a href="https://www.popsci.com/technology/digital-haunting-chatbots/">Their scenarios</a> are unsettling precisely because they are so plausible: a dead grandparent’s voice used to sell you things, a child who keeps talking to a simulation of a parent who has died, a family locked into a subscription they cannot cancel.</p><p>“These services run the risk of causing huge distress to people,” Hollanek warned. The researchers called for ways to retire a digital copy with dignity, for clear consent from the person being copied, and for keeping these services away from children.</p><p>A photograph does not text you on your birthday. A digital copy can. That is its power, and its danger.</p><h3>So, would it be you?</h3><p>The honest answer is: probably not. At least, not in the way you most want it to be.</p><p>When people ask this question, they are rarely asking about metaphysics. They are asking whether they have to die. Whether the person who lies awake at three in the morning, who loves their children, who is reading this sentence, gets to keep going. On every serious account, a copy does not give them that. Something like you would wake up. You would not.</p><p>But that is not the end of the story, and Parfit knew it. Working through these cases changed how he felt about his own death. “My life seemed like a glass tunnel,” he <a href="https://www.lrb.co.uk/blog/2017/january/remembering-derek-parfit">wrote</a>, “through which I was moving faster every year, and at the end of which there was darkness. When I changed my view, the walls of my glass tunnel disappeared. I now live in the open air.”</p><p>What he meant was this. If identity is not the thing that matters, death stops being an absolute wall. What matters is connection: how much of you carries forward into the people and the world you leave behind. And that, unlike identity, is something we can actually do something about.</p><p>A digital copy will not be you. But it might carry some of you. Your stories. Your way of explaining things. The advice you would have given your granddaughter on the day she needed it most. That is not immortality. It might be the next best thing: being remembered accurately, in your own words, by the people who loved you.</p><p>At Sensay, we work on a smaller, more practical version of this every day: capturing what people know before it is lost. The longer we do it, the clearer one thing becomes. Nobody can preserve a whole person. But almost everyone is carrying something worth keeping, and almost none of it is written down.</p><p>So perhaps the real question is not whether a copy of you would be you. It is what you would want it to say.</p><h3>Sources</h3><ul><li><a href="https://gwern.net/doc/www/www.sfchronicle.com/70bbc07b9a27d32fd8b69bba025aceee6628bd66.html">The Jessica Simulation</a>, Jason Fagone, San Francisco Chronicle, July 2021 (archived copy)</li><li><a href="http://www.theverge.com/a/luka-artificial-intelligence-memorial-roman-mazurenko-bot">Speak, Memory</a>, Casey Newton, The Verge, October 2016</li><li><a href="https://www.popsci.com/technology/digital-haunting-chatbots/">AI ethicists warn of ‘digital hauntings’ from dead loved ones</a>, Popular Science, May 2024, reporting Hollanek and Nowaczyk-Basińska in <em>Philosophy &amp; Technology</em></li><li><a href="https://www.lrb.co.uk/blog/2017/january/remembering-derek-parfit">Remembering Derek Parfit</a>, London Review of Books blog, January 2017, quoting <em>Reasons and Persons</em> (1984)</li><li><a href="https://personal.lse.ac.uk/robert49/teaching/ph103/lecture5.html">Ph103 Lecture 5: Personal identity</a>, London School of Economics, on Parfit’s teletransporter and branch-line cases</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6b4d02491d6b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Hidden Cost of Losing Institutional Memory]]></title>
            <link>https://asksensay.medium.com/the-hidden-cost-of-losing-institutional-memory-99b72ab4d34e?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/99b72ab4d34e</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Wed, 04 Nov 2026 19:14:16 GMT</pubDate>
            <atom:updated>2026-11-04T19:14:16.739Z</atom:updated>
            <content:encoded><![CDATA[<p>Nov 4, 2026 · @Dan Thomson</p><p>In the 1980s, the United States made a classified material called Fogbank for its W76 nuclear warheads. Then it stopped. The facilities closed in the 1990s.</p><p>When the warheads needed refurbishing in the 2000s, the National Nuclear Security Administration discovered it could no longer make the stuff. According to the <a href="https://webarchive.library.unt.edu/web/20130214015304/http://www.gao.gov/new.items/d09385.pdf">US Government Accountability Office</a>, the agency “had kept few records of the process”, and “almost all staff with expertise on production had retired or left the agency.”</p><p>The programme slipped by a year, and the effort fed into a $69 million cost overrun. Nobody had lost a file. They had lost the people who knew how it was actually done.</p><p>Most companies are not making warhead components. But nearly all of them are quietly running the same experiment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3Spc6-oFjPG7CjDiZLo67A.png" /></figure><h3>Why the cost stays hidden</h3><p>When a senior person leaves, the visible costs get counted: recruitment fees, a signing bonus, a few weeks of an empty desk. The cost of what they knew rarely appears anywhere. There are three reasons.</p><p><strong>There is no line item.</strong> No budget has a row called “things only Priya knew”. The loss is spread across dozens of small delays and mistakes, none big enough to trace back.</p><p><strong>It arrives late.</strong> The day someone leaves, nothing breaks. The damage shows up months later, when a renewal comes round, an annual process runs or a system fails in a way it last failed four years ago.</p><p><strong>It gets blamed on something else.</strong> A deal slips and it looks like a sales problem. A release goes wrong and it looks like a quality problem. A new hire struggles and it looks like a hiring problem. The real cause is that the answer left the building, and that rarely makes it into the post-mortem.</p><h3>Where the cost actually shows up</h3><p><a href="https://www.gallup.com/workplace/247391/fixable-problem-costs-businesses-trillion.aspx">Gallup</a> estimates that replacing an employee costs between half and twice their annual salary, and calls that conservative. Most people read that number as recruiting and training. The knowledge cost is harder to see. It tends to surface in six places.</p><p><strong>1. Slower ramp-up.</strong> A successor has to rebuild context the last person carried in their head. In a <a href="https://www.hrdive.com/news/inefficient-knowledge-sharing-costs-large-us-businesses-47m-a-year/527892/">Panopto survey reported by HR Dive</a>, new hires were formally trained for about 2.5 months but could take as long as six months to get up to speed.</p><p><strong>2. Repeated mistakes.</strong> Lessons learned the hard way get learned the hard way again. The supplier that let you down, the migration that corrupted data, the clause that caused a dispute. Without the person who remembers, the organisation pays for the same lesson twice.</p><p><strong>3. Relitigated decisions.</strong> When nobody remembers why a decision was made, it gets reopened. Teams spend weeks debating something that was settled years ago, or quietly reverse a good decision because its reasoning was never written down.</p><p><strong>4. Customer relationships.</strong> Account knowledge is rarely in the CRM. Who the real decision-maker is, what went wrong two renewals ago, which promises were made over dinner. When the account manager leaves, the customer notices before the company does.</p><p><strong>5. Risk and compliance.</strong> Controls outlive the people who designed them. When nobody knows why a check exists, it gets skipped or removed, and the risk it was guarding against comes back.</p><p><strong>6. Knock-on attrition.</strong> The colleagues left behind become the help desk. The same Panopto survey found that 65% of employees in high-turnover companies found it difficult or nearly impossible to get the information they needed to do their jobs. Frustration like that is how one departure turns into three.</p><h3>A rough way to put a number on it</h3><p>You will never get a precise figure, but you can get a useful one. Pick one person whose departure would worry you, and answer four questions.</p><ol><li><strong>How long would a successor take to reach full speed?</strong> Be honest. For senior roles it is usually months, not weeks.</li><li><strong>How productive are they in the meantime?</strong> Half is a reasonable working assumption for most knowledge roles.</li><li><strong>What did the last “only they knew” problem cost?</strong> A delayed launch, a lost renewal, a week of engineers chasing a fault someone else would have spotted in an hour.</li><li><strong>How likely are they to leave in the next year?</strong> Retirement plans, market demand for their skills, how long since their last promotion.</li></ol><p>A worked example, with illustrative numbers. A senior engineer earns $150,000 a year. Their successor takes six months to reach full speed, working at half productivity in the meantime. That is about $37,500 of lost output before anything goes wrong.</p><p>That is the floor, not the estimate. In most organisations, question three is where the real money is. One lost renewal or one long outage usually costs more than the entire ramp-up.</p><h3>Why it is getting more expensive</h3><p>The usual story is that job-hopping is draining companies of knowledge. The data says something more specific.</p><p>According to the <a href="https://www.bls.gov/news.release/archives/tenure_09242026.htm">US Bureau of Labor Statistics</a>, median tenure with a current employer was 4.1 years in January 2026, up slightly from 3.9 years in 2024. Overall churn is not the headline. Concentration is. Workers aged 55 to 64 had a median tenure of 9.6 years, more than three times the 3.0 years of workers aged 25 to 34.</p><p>In other words, the deepest institutional memory sits with the people closest to retirement, and that group is reaching retirement age in record numbers. The <a href="https://www.protectedincome.org/wp-content/uploads/2024/01/Peak-65-Release-final-011024.pdf">Alliance for Lifetime Income</a> estimates more than 4.1 million Americans will turn 65 each year to 2027.</p><p>Remote and hybrid work adds a quieter problem. Much institutional memory used to pass on by accident: overheard calls, questions at the next desk, the story told over lunch. Fewer of those moments happen now, so less of it passes on before people go.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IAFQMcul8eaSaw9WXij-8w.png" /></figure><h3>What to do about it</h3><p>You cannot stop people leaving. You can stop their knowledge leaving with them.</p><ul><li><strong>Find the single points of failure.</strong> List the people whose absence would stall something important. Most leadership teams can name them in ten minutes. Few have ever written the list down.</li><li><strong>Capture continuously, not at exit.</strong> By the notice period it is too late to do properly. Make knowledge capture a routine part of critical roles, while people are engaged and have time.</li><li><strong>Capture judgement, not just process.</strong> The steps are the easy part. The value is in the cues, exceptions and trade-offs. We covered how to draw those out in <a href="https://asksensay.medium.com/how-to-interview-for-judgement-not-just-instructions-45e2bb51aa02">How to interview for judgement, not just instructions</a>.</li><li><strong>Put it where people already ask questions.</strong> Knowledge in a wiki nobody opens has the same effect as knowledge nobody kept. It needs to be answerable in the tools people use every day.</li><li><strong>Count the incidents.</strong> Every time something stalls because “only they knew”, log it. A few months of that list makes the hidden cost visible, which is usually what it takes to get budget.</li></ul><p>This is the problem Sensay was built to solve. It runs AI-guided interviews with the people who hold critical knowledge, then makes what they know available as answers in Slack and Microsoft Teams. People still move on. Their answer to “why do we do it this way?” stays.</p><h3>The cheapest time to act</h3><p>The NNSA did not set out to forget how to make Fogbank. It simply stopped needing the knowledge for a while, and let the people who held it go. By the time it needed them again, they were gone.</p><p>Every organisation has its own version of that, waiting for the moment someone asks a question nobody can answer. The cheapest time to capture that knowledge is while the person who holds it is still at their desk.</p><p>If you want to see how Sensay does that, <a href="https://sensay.io/">get in touch</a>.</p><h3>Sources</h3><ul><li><a href="https://webarchive.library.unt.edu/web/20130214015304/http://www.gao.gov/new.items/d09385.pdf">Nuclear Weapons: NNSA and DOD Need to More Effectively Manage the Stockpile Life Extension Program</a>, US Government Accountability Office, GAO-09–385, March 2009</li><li><a href="https://www.gallup.com/workplace/247391/fixable-problem-costs-businesses-trillion.aspx">This Fixable Problem Costs U.S. Businesses $1 Trillion</a>, Gallup, March 2019</li><li><a href="https://www.hrdive.com/news/inefficient-knowledge-sharing-costs-large-us-businesses-47m-a-year/527892/">Inefficient knowledge-sharing costs large US businesses $47M a year</a>, HR Dive, July 2018 (reporting the Panopto Workplace Knowledge and Productivity Report)</li><li><a href="https://www.bls.gov/news.release/archives/tenure_09242026.htm">Employee Tenure in January 2026</a>, US Bureau of Labor Statistics, September 2026</li><li><a href="https://www.protectedincome.org/wp-content/uploads/2024/01/Peak-65-Release-final-011024.pdf">The Peak 65 Zone is Here</a>, Alliance for Lifetime Income, January 2024</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=99b72ab4d34e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to interview for judgement, not just instructions]]></title>
            <link>https://asksensay.medium.com/how-to-interview-for-judgement-not-just-instructions-45e2bb51aa02?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/45e2bb51aa02</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Tue, 20 Oct 2026 17:45:12 GMT</pubDate>
            <atom:updated>2026-10-20T17:45:12.018Z</atom:updated>
            <content:encoded><![CDATA[<p>Ask a senior engineer how they handle a production incident and you will get a tidy answer. Check the alerts, find the failing service, roll back, write it up. All true. None of it is why they are good at it.</p><p>What makes them good is the bit they skip. They glance at one graph and know it is a database problem, not a network one. They know which alerts to ignore on a Monday morning. They know when to wake the CTO and when to sit on it for twenty minutes.</p><p>That is judgement. It is the most valuable thing an expert carries, and the standard knowledge-capture interview is almost designed to miss it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UP1VMKV_9G0y2nI2V6pN6Q.png" /></figure><h3>Instructions are not the job</h3><p>Instructions tell you what to do when things go to plan. Judgement tells you what to do when they don’t, which is most of what senior people are paid for.</p><p>When you interview someone before they leave, these are the things you are actually after:</p><ul><li><strong>Cues.</strong> What they notice that others miss. The tone of an email, a number that looks slightly off, a client who has gone quiet.</li><li><strong>Exceptions.</strong> When the rule does not apply, and how they know.</li><li><strong>Trade-offs.</strong> What they sacrifice, and when. Speed or accuracy, the relationship or the margin.</li><li><strong>Mental models.</strong> Their picture of how the system really behaves, which is rarely the one in the documentation.</li><li><strong>Anticipation.</strong> What they expect to happen next, so they act before it does.</li></ul><p>None of these is a step in a process, so none of them turns up when you ask for the process.</p><h3>Why the obvious questions fail</h3><p>The philosopher Michael Polanyi put it neatly in <a href="https://en.wikipedia.org/wiki/Tacit_knowledge">The Tacit Dimension</a> (1966): “we can know more than we can tell”. Researchers who study expert knowledge have since tried to measure the gap. <a href="https://digitalcommons.usu.edu/itls_facpub/454">Yates, Feldon and Clark</a>, summarising their work on cognitive task analysis, put it at about 70% of the decisions experts make when they describe how to do a complex task.</p><p>This is not evasiveness. Expertise becomes automatic. The more often someone has made a call, the less it feels like a call, so it drops out of their account.</p><p>That is why the standard offboarding questions disappoint. “Talk me through your process” gets you the process. “What should your successor know?” gets you whatever is front of mind on a busy Thursday. Both questions ask the expert to do the hard part, spotting their own tacit knowledge, which is exactly what they cannot do unprompted.</p><h3>Six techniques that get to the judgement</h3><p>Most of what works comes from cognitive task analysis, in particular the <a href="https://www.gary-klein.com/cdm">Critical Decision Method</a> that Gary Klein and colleagues developed in the late 1980s with fire ground commanders. It has since been used to capture expertise in healthcare, the military and emergency response. You do not need the full academic method. You need its core moves.</p><ol><li><strong>Anchor on a real incident.</strong> Never ask how they usually do something. Ask about one specific time it got difficult. “Tell me about the last renewal that nearly fell through.” Real events carry detail that generalisations flatten out, and a specific memory is much harder to tidy up than a general description.</li><li><strong>Go through it more than once.</strong> First pass: let them tell the story. Second pass: build the timeline and mark the moments where they had to decide something. Third pass: slow down at each of those moments. The judgement sits in the third pass.</li><li><strong>Ask what they noticed.</strong> “What told you something was off?” “What were you looking at?” “What would you have expected to see if it was fine?” Cues are the part of expertise people are least aware of, and the part successors most need.</li><li><strong>Ask what a newcomer would get wrong.</strong> “If someone six months into the role had been handling this, where would they have slipped?” Experts struggle to describe their own skill. They are very good at spotting its absence.</li><li><strong>Change the facts.</strong> “What if it had been a new client?” “What if the deadline was real?” “What would have made you do the opposite?” Hypotheticals show where the decision boundaries are, which a single story cannot.</li><li><strong>Ask when they break the rule.</strong> “Which parts of the process do you skip, and when?” Every experienced person has a list. It is usually the most useful thing they will tell you, and the thing they are least likely to volunteer.</li></ol><h3>Swap the question</h3><p>Most weak questions can be fixed by making them specific, past tense and about one real situation.</p><ol><li><strong>“Walk me through the last time this went wrong.”</strong> Not “Talk me through your process.” It brings out decision points rather than steps.</li><li><strong>“What did you learn the hard way in your first year?”</strong> Not “What should your successor know?” It brings out lessons and near misses.</li><li><strong>“Last Monday, what did you deal with first, and why?”</strong> Not “How do you prioritise?” It brings out the real trade-offs.</li><li><strong>“Who do you go to when the official route stalls?”</strong> Not “Who are the key contacts?” It brings out the real org chart.</li><li><strong>“Which steps do you skip, and when?”</strong> Not “Are there any exceptions?” It brings out where the rules bend.</li><li><strong>“What do you check that is not in the document?”</strong> Not “Is it all documented?” It brings out the workarounds.</li><li><strong>“What would worry you if you saw it on a dashboard?”</strong> Not “How do you know it is working?” It brings out cues and warning signs.</li></ol><h3>Running the session</h3><p>Good questions only work if the session gives them room.</p><ul><li><strong>Start early.</strong> Interview people while they are still engaged, not in their last week. Better still, make it routine for anyone in a critical role, leaving or not.</li><li><strong>Keep sessions short and come back.</strong> An hour is plenty. People remember more between sessions, and the second conversation is usually richer than the first.</li><li><strong>Prepare incidents, not topics.</strong> Before the session, ask the expert to think of two or three situations that were genuinely hard. That alone shifts the conversation from process to judgement.</li><li><strong>Follow the thread.</strong> When they say “it depends”, stop there. Ask what it depends on. That phrase is where tacit knowledge is hiding.</li><li><strong>Record it.</strong> Notes capture conclusions. Recordings capture the reasoning, the hesitations and the asides, which is often where the real answer sits.</li><li><strong>Play it back.</strong> Summarise what you heard and ask what you got wrong. Experts are much better at correcting an account than producing one from scratch.</li><li><strong>Include the successor.</strong> Where possible, let the person who inherits the role ask their own questions. They know what they do not know better than anyone.</li></ul><h3>Doing this at scale</h3><p>None of this needs software. A thoughtful manager with an hour and the right questions will get more out of a departing expert than a 12-page template ever will.</p><p>The problem is that it rarely happens. Good interviewing takes time and skill, and it competes with everything else in someone’s last month. So it gets skipped, or squeezed into a single rushed conversation.</p><p>That is the gap Sensay is built for. Our AI-guided interviews follow up on the vague answers, come back to the “it depends” and work around the expert’s calendar rather than a manager’s. What they capture becomes answers colleagues can ask for in Slack and Microsoft Teams, long after the expert has moved on.</p><h3>Start with one question</h3><p>Pick the person whose departure would worry you most. Next time you speak, skip “how do you do this?” and ask instead: “Tell me about the last time this got difficult. What did you notice first?”</p><p>Then listen for the part they did not plan to tell you.</p><p>If you want to see how Sensay captures that judgement before it walks out of the door, <a href="https://sensay.io/">get in touch</a>.</p><h3>Sources</h3><ul><li><a href="https://en.wikipedia.org/wiki/Tacit_knowledge">Tacit knowledge</a>, Wikipedia, citing Michael Polanyi, The Tacit Dimension (University of Chicago Press, 1966)</li><li><a href="https://digitalcommons.usu.edu/itls_facpub/454">Using cognitive task analysis to capture expert knowledge and skills for research and instructional design</a>, Yates, Feldon and Clark, AERA workshop, 2011</li><li><a href="https://www.gary-klein.com/cdm">Critical Decision Method</a>, Gary Klein</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=45e2bb51aa02" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The knowledge that never makes it into a handover]]></title>
            <link>https://asksensay.medium.com/the-knowledge-that-never-makes-it-into-a-handover-109be7e312f0?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/109be7e312f0</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Sat, 17 Oct 2026 17:07:06 GMT</pubDate>
            <atom:updated>2026-10-17T17:07:06.740Z</atom:updated>
            <content:encoded><![CDATA[<p>The handover document is usually fine. Twelve pages, a shared drive link, a list of open projects, the password manager entries and a warm note about who to call. Everyone signs it off in the final week.</p><p>Then, about three weeks later, someone asks why one client is always invoiced on the 3rd rather than the 1st. Nobody knows. The person who knew has gone, and the document never mentioned it, because to them it was never information. It was just how things worked.</p><p>This is the knowledge that never makes it into a handover. It is usually the part that mattered most.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HqnrYrXGBhMeK9giurtDGA.png" /></figure><h3>What a handover actually captures</h3><p>Handovers are good at the explicit layer: what exists, where it lives and what is due next. Process steps, system access, project status, a contact list. This is knowledge the leaver knows they have, and can write down in an afternoon.</p><p>What they miss is the layer underneath. Why things are the way they are, and what to do when the process runs out. Academics call it tacit knowledge. Everyone else calls it “how things actually work round here”.</p><p>It rarely gets written down because nobody experiences it as information. It feels like common sense, right up until the person with the common sense leaves.</p><h3>Five things that never make it in</h3><ol><li><strong>The why.</strong> The reason a process has an odd extra step. The reason the team stopped using a particular supplier. Without it, the successor either follows rules blindly or “fixes” something that was built that way on purpose.</li><li><strong>The who.</strong> Not the org chart, the real one. Who actually moves things in finance. Which customer contact to try when the official one goes quiet. Whose opinion matters even though their sign-off is not required.</li><li><strong>The workarounds.</strong> The spreadsheet that patches a gap in the CRM. The manual check before the monthly report goes out, because the export double-counts refunds. Workarounds are invisible by design, which is exactly why they vanish.</li><li><strong>The near misses.</strong> What almost went wrong last year, and what was learned from it. Organisations remember their disasters. They forget their near misses, which were the cheaper lessons.</li><li><strong>The judgement calls.</strong> When to escalate and when to sit tight. Which deadlines are real. How hard to push a particular client. This is the hardest knowledge to write down and the most expensive to relearn.</li></ol><h3>Why handovers miss it</h3><p>Leavers are rarely careless. The format sets them up to fail, for three reasons.</p><p><strong>People do not know what they know.</strong> Ask an expert to write down what they do and you get the procedure. Ask what they would do if a specific thing went wrong and you get the expertise. The second only comes out in answer to questions.</p><p><strong>The timing is wrong.</strong> Handovers happen during notice periods, when the leaver is wrapping up, mentally half out of the door and usually still doing the day job. The people with the most to hand over tend to have the least time to do it.</p><p><strong>The format is wrong.</strong> A document answers the questions its author thought of. The successor’s real questions arrive weeks or months later, one at a time, in the middle of a task. By then the only person who could answer them has gone.</p><h3>Why this is getting more expensive</h3><p>In a <a href="https://www.hrdive.com/news/inefficient-knowledge-sharing-costs-large-us-businesses-47m-a-year/527892/">2018 Panopto survey</a> of more than 1,000 US workers, respondents said 42% of the knowledge they need to do their job is unique to them and not shared with colleagues. The same study found knowledge workers lose 5.3 hours a week waiting for information or recreating what already exists. It is self-reported, so treat it as directional. It matches what most managers see.</p><p>The demographics make it sharper. The <a href="https://www.protectedincome.org/wp-content/uploads/2024/01/Peak-65-Release-final-011024.pdf">Alliance for Lifetime Income</a> estimates more than 4.1 million Americans will turn 65 each year from 2024 to 2027, the largest such wave in US history. Many of them hold exactly the long-tenure, relationship-heavy knowledge that handovers miss.</p><p>When a 30-year veteran retires, the handover document is not short of pages. It is short of 30 years.</p><h3>What works instead</h3><p>The fix is not a better template. It is a different method.</p><ul><li><strong>Interview, do not ask for a write-up.</strong> Tacit knowledge comes out under questioning. “What would you do if this broke?” “Why does this step exist?” “Who do you call when the official route stalls?” A good interviewer follows the thread the leaver did not know was there.</li><li><strong>Start before the notice period.</strong> Capture goes better when it is not squeezed into someone’s last fortnight. The best time to record what a senior person knows is while they are still enjoying the job.</li><li><strong>Make it answerable, not readable.</strong> Nobody reads a 12-page handover cover to cover. They have one specific question on a Tuesday afternoon. The knowledge needs to be askable in the tools where that question comes up.</li></ul><p>This is what we built Sensay to do. It runs AI-guided interviews with the people who hold critical knowledge, then makes what they said available to colleagues as answers inside Slack and Microsoft Teams. The person still leaves. Their answer to “why do we do it this way?” stays.</p><h3>One question before the next leaving drinks</h3><p>Every organisation has someone whose departure would quietly break things that appear on no process map. Work out who that is in yours, then write down what you would need to ask them. If you cannot answer the second part, their handover will not answer it either.</p><p>If you want to see how Sensay captures that knowledge before it walks out of the door, <a href="https://sensay.io/">get in touch</a>.</p><h3>Sources</h3><ul><li><a href="https://www.hrdive.com/news/inefficient-knowledge-sharing-costs-large-us-businesses-47m-a-year/527892/">Inefficient knowledge-sharing costs large US businesses $47M a year</a>, HR Dive, July 2018 (reporting the Panopto Workplace Knowledge and Productivity Report)</li><li><a href="https://www.protectedincome.org/wp-content/uploads/2024/01/Peak-65-Release-final-011024.pdf">The Peak 65 Zone is Here</a>, Alliance for Lifetime Income, January 2024</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=109be7e312f0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Social Media Is Dead. Nobody Told the Bots.]]></title>
            <link>https://asksensay.medium.com/social-media-is-dead-nobody-told-the-bots-04f5c564f967?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/04f5c564f967</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Tue, 06 Oct 2026 13:47:24 GMT</pubDate>
            <atom:updated>2026-10-06T13:47:24.912Z</atom:updated>
            <content:encoded><![CDATA[<p>Oct 6, 2026 · Dan Thomson</p><p><em>The death of social media for businesses, and why the bots will be the last to notice.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wyULsPWTiiiwPxIFEaFWZQ.png" /></figure><p>Right now, an AI is writing a LinkedIn post about the importance of authenticity. A second AI is liking it. A third is leaving the comment “Really valuable insight, thanks for sharing!” and a fourth is sliding into the author’s DMs offering to triple their engagement. No human will read any of it, and the analytics dashboard will call it a good day.</p><p>I build AI for a living, so this isn’t a Luddite talking. But I’ve reached a conclusion that should worry anyone who owns a content calendar: for businesses, social media is finished. The feeds still scroll and the numbers still climb. What has gone is the one thing that ever made it worth doing, which is that people believed what they saw there.</p><h3>The feed is already half machine</h3><p>Start with what you scroll. In July, the AI detection firm Pangram published data from roughly a million posts its users were actually served over two months across LinkedIn, X, Reddit, Substack and Medium. On LinkedIn, <a href="https://404media.co/linkedin-and-x-are-flooded-with-ai-spam-browsing-data-suggests">41% of long-form posts were fully AI-generated</a>. On X it was 25%, with another 23% AI-assisted. Pangram’s own chief executive calls those figures a lower bound.</p><p>Zoom out and it gets stranger. Imperva’s <a href="https://community.imperva.com/blogs/tim-chang/2026/05/05/the-2026-bad-bot-report-is-here">2026 Bad Bot Report</a> puts automated traffic at 53% of all web traffic, the second year running that machines have out-numbered people. Even Sam Altman, whose company did more than anyone to make this possible, <a href="https://futurism.com/sam-altman-dead-internet-theory">admitted last year</a> that there now seem to be “a lot of LLM-run twitter accounts”.</p><p>Before anyone sends me a correction: no, not everything online is slop. Plenty of feeds are still mostly human, and detection tools get things wrong. But that misses the point. You don’t need a feed to be 100% fake to break a system built on trust. You only need enough fake that nobody can tell which bits are real, and we crossed that line a while ago.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*axMNcTngwPRSlCzlmpNFmQ.png" /></figure><h3>Then the replies and the DMs went</h3><p>The comments under a post used to be the proof it had landed. Now they are where the automation hides. “Great insights!” “So true!” “Thanks for sharing, Sarah!” Most of us can spot them, and businesses buy them anyway, because a busy comment section tells the algorithm a post deserves more reach.</p><p>To be fair to the data, replies are still the most human corner of the feed. Pangram found original posts were more likely to be machine-written than replies, and on Reddit <a href="https://gigazine.net/gsc_news/en/20260714-linkedin-longform-content-ai/">about 98% of replies looked human</a>. I wouldn’t bank on that lasting. A reply is cheaper to automate than a post, and the incentive is exactly the same.</p><p>The inbox has already gone. Anyone with a LinkedIn profile now gets a daily stream of “personalised” messages that mention their latest post, compliment their company and pitch them something, all written by a sales tool that has never heard of them.</p><p>And it isn’t only noise. In 2024, a finance worker at Arup in Hong Kong <a href="https://edition.cnn.com/2024/05/16/tech/arup-deepfake-scam-loss-hong-kong-intl-hnk">transferred around US$25.6 million</a> to fraudsters after a video call in which the CFO and several colleagues were all deepfakes. If a live call with your own CFO can be faked, a DM from a “prospective customer” could be anything.</p><h3>Why the machines beat us at the algorithm</h3><p>Here is the uncomfortable bit. Social algorithms don’t reward truth or quality. They reward signals: posting frequency, early engagement, dwell time, the shape of a hook. Those are patterns, and finding patterns at scale is the one thing AI does better than any human team.</p><p>A good marketer can write three strong posts a week. A model can write three hundred versions of a hook before lunch, test them, read what the platform rewards and rewrite to suit. It never gets tired, never runs dry and never posts at the wrong hour. A human social team competing with that is a rowing crew entered in a powerboat race.</p><p>The data shows where this bites. An <a href="https://originality.ai/blog/linkedin-ai-study-engagement">Originality.ai study</a> of LinkedIn posts from 99 influential profiles found humans still won on engagement in most sectors. But in leadership and inspiration, the most formulaic genre on the platform, AI-written posts averaged 75% more engagement than human ones. Most business content is formulaic. That is precisely the ground the machines are taking.</p><p>Now the platforms are joining in. Meta’s stated goal, <a href="https://www.verdict.co.uk/meta-automate-ad-creation/">reported by the Wall Street Journal</a>, is that by the end of 2026 a brand can upload a product image and a budget and let Meta’s AI write the ad, make the visuals, pick the audience and decide where the money goes. At that point the business isn’t really marketing. It’s paying one machine to talk to people who increasingly assume they’re talking to a machine.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_8tqpjbmeiPjm6PGehZwSA.png" /></figure><h3>The real casualty is trust</h3><p>The obvious damage is noise. The worse damage is what the noise does to everything genuine sitting next to it.</p><p>In 2018 the legal scholars Bobby Chesney and Danielle Citron gave this a name: the liar’s dividend. Once people know convincing fakes exist, anything real can be waved away as one. Social media now runs on the commercial version. Your customer case study, your founder’s honest post about a hard quarter, the glowing note from a real client: all of it takes the same discount as the slop around it, because the reader has no way to tell them apart.</p><p>That quietly breaks every tool businesses relied on. Followers can be bought. Engagement can be generated. Reviews can be written by the thousand. Social proof only works when the proof is hard to fake, and none of it is any more.</p><p>My bet is that buyers have already adjusted, even if most marketing teams haven’t. They go back to the things that are expensive to fake: a recommendation from someone they know, a private group chat, a conversation with a person they have actually met.</p><h3>What businesses should do instead</h3><p>None of this means going dark. It means accepting that broadcasting to strangers through someone else’s algorithm has stopped working, and moving effort to where trust still lives.</p><p><strong>Prove you’re human, and make it costly.</strong> Anything cheap to produce is now assumed to be fake, so the things that carry weight are the things that are hard to fake: live sessions where people can ask awkward questions, events in real rooms, your actual people on camera saying things a model didn’t draft.</p><p><strong>Own the relationship.</strong> An email list, a community you run, a customer you can ring. No algorithm can throttle these, and bots struggle to flood them.</p><p><strong>Publish what only you know.</strong> AI can produce endless generic content because generic content is what it was trained on. It cannot produce your hard-won specifics: the project that went wrong and what it taught you, the customer story with real numbers, the judgement of the engineer who has been doing this for thirty years. That tacit knowledge is the scarcest thing left online. Put it out under real names.</p><p><strong>Use AI for the plumbing, not the voice.</strong> Research, scheduling, analytics and first drafts nobody else sees are fair game. The words your customers read should come from someone they could meet.</p><p><strong>Change what you measure.</strong> Impressions and likes now partly count how many machines saw your post. Count replies from people you can name, meetings booked and deals that began with “I saw what you wrote”.</p><h3>The feed will keep scrolling</h3><p>Social media isn’t going to vanish. The platforms will keep growing, the dashboards will keep glowing and the bots will keep posting to each other long after the last real buyer has stopped looking. That is what makes it dangerous for a business: it will go on looking like it works.</p><p>The winners of the next five years will be the companies people still believe. Very little of that will be decided in a feed.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=04f5c564f967" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Send Sensay (SNSY) Mass Payouts with ChangeNOW PRO]]></title>
            <link>https://asksensay.medium.com/how-to-send-sensay-snsy-mass-payouts-with-changenow-pro-c497c5a54bf6?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/c497c5a54bf6</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Mon, 31 Aug 2026 16:36:01 GMT</pubDate>
            <atom:updated>2026-08-31T16:36:01.200Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*br5G5k6LwvIkFLigUviOIg.png" /></figure><p>In this article, you’ll learn:</p><ul><li>Why Sensay is suitable for recurring business payouts.</li><li>What makes SNSY payouts difficult to manage at scale.</li><li>How NOWPayments supports payments, payouts, conversions, and treasury operations.</li><li>How to send SNSY by email through ChangeNOW PRO in under one second with zero network fees.</li><li>How to set up the payout workflow with NOWPayments.</li></ul><p><a href="https://snsy.ai/">Sensay</a> is an AI ecosystem built around digital replicas and AI-powered knowledge management applications, and SNSY is increasingly used for affiliate commissions, payroll, creator rewards, marketplace settlements, gaming prizes, and community payments. Multi-chain availability across Ethereum, Base, and Arbitrum, growing adoption by payment gateways and exchanges, and clear utility within the Sensay platform make SNSY a practical asset for recurring business payouts.</p><p>The challenge begins when payout volumes grow. Collecting wallet addresses, validating recipient details, tracking transactions, and reconciling payments can create significant manual work. <a href="https://nowpayments.io/">NOWPayments</a> brings payment acceptance, mass payouts, conversions, treasury workflows, enterprise APIs, and operational support into one crypto business ecosystem. Through ChangeNOW PRO, businesses can also send SNSY using recipient email addresses instead of wallet addresses. This guide explains why companies choose SNSY and how to manage the payout process with NOWPayments.</p><h3>Why SNSY Has Become a Practical Choice for Business Payouts</h3><p>Different cryptocurrencies serve different business needs. Some companies choose stablecoins for price stability, while others prefer ecosystem tokens with real utility, broad infrastructure support, and strong integration with payment rails. Sensay (SNSY) belongs to the second category.</p><ul><li>Multi-chain availability. SNSY is deployed as an ERC-20 token on Ethereum and as an Omnichain Fungible Token (OFT) on Base and Arbitrum through LayerZero, giving businesses flexibility in how they move and settle funds.</li><li>Growing ecosystem integrations. SNSY is accepted for travel bookings through Travala, integrated with NOWPayments for merchant payments, and listed on exchanges and swap platforms, including ChangeNOW.</li><li>Utility within the Sensay platform. Businesses and users can use SNSY for subscriptions, premium features, and services inside the Sensay AI ecosystem.</li><li>Suitable for recurring payouts. Businesses can use SNSY for affiliate commissions, payroll, creator rewards, marketplace settlements, and gaming payouts.</li><li>Suitable for upcoming user payouts from enterprise clients.</li></ul><p>While Sensay provides the token and ecosystem, businesses still need an efficient way to organize and deliver payouts at scale.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7BqSfLi3uMBAcn_pu6tqZg.png" /></figure><h3>The Real Challenge Begins After You Decide to Use SNSY</h3><p>Sending SNSY to a handful of recipients is straightforward. As payout volumes grow, however, the challenge shifts from the token itself to the business operations required to manage recurring payments at scale.</p><p>Every large payout program involves much more than sending transactions.</p><p><strong>Payout Workflow &amp; Operational Responsibility</strong></p><p><strong>Wallet Collection</strong></p><p>Gather and maintain accurate recipient ERC-20 addresses on Ethereum, Base, or Arbitrum.</p><p><strong>Address Validation</strong></p><p>Verify wallet details and network compatibility before every payout.</p><p><strong>Transaction Monitoring</strong></p><p>Confirm that each payout has been completed successfully.</p><p><strong>Payment Reconciliation</strong></p><p>Match blockchain transactions with accounting records and invoices.</p><p><strong>Recipient Support</strong></p><p>Handle payment inquiries and wallet update requests.</p><p>None of these challenges come from Sensay itself. They appear when blockchain transactions become part of a growing business. That’s where dedicated payout infrastructure becomes just as important as the token powering the transaction.</p><h3>Building SNSY Payout Infrastructure with NOWPayments</h3><p>Accepting crypto is only one part of running a digital asset business. As operations grow, companies also need payouts, treasury, automation, and settlement infrastructure. Managing each of these through separate providers can quickly increase operational complexity.</p><p>NOWPayments brings these capabilities together in one crypto business ecosystem:</p><ul><li><strong>Payments.</strong> Accept major cryptocurrencies and stablecoins, including USDT and USDC, with fees from 0.3% for eligible partners.</li><li><strong>Mass payouts.</strong> Automate high-volume crypto distributions with average delivery in 1–3 minutes or under one second through ecosystem email payouts.</li><li><strong>Conversions &amp; treasury.</strong> Convert assets with a 0.5% conversion fee and manage balances and settlements from one place.</li><li><strong>Enterprise APIs.</strong> Connect payments, payouts, and treasury workflows through a single integration.</li><li><strong>Business support.</strong> Scale with enterprise onboarding, dedicated account management, 24/7 support, and operational guidance.</li></ul><p>One of the latest additions to this ecosystem is <a href="https://nowpaymentsinteractiveguide.pages.dev/"><strong>Zero-Fee Ecosystem Payouts</strong></a> through ChangeNOW PRO. Businesses can send SNSY using a recipient’s email address instead of a wallet address, making large-scale payouts faster and easier to manage. <a href="https://zerofeepayouts.netlify.app/"><strong>Learn more about Zero-Fee Ecosystem Payouts</strong></a> and how the workflow works.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vKWCQJzmTIkvJgUxS0KVJw.png" /></figure><h3>How to Send Crypto Mass Payouts in Sensay (SNSY) via ChangeNOW PRO</h3><p>Traditional crypto payouts require businesses to collect wallet addresses. With ChangeNOW PRO, recipients can be identified by email instead.</p><ol><li><strong>Access your account.<br> </strong><a href="https://account.nowpayments.io/sign-in">Sign up or log in</a> to NOWPayments.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HC-fQUHUYvdYsVduKk9SDQ.png" /></figure><p><strong>2. Open </strong><a href="https://account.nowpayments.io/mass-payouts"><strong>Mass Payouts</strong></a><strong>.<br> </strong>Use the dashboard or the existing API endpoint.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uTgs7MX3sJJxOID_3-NNNQ.png" /></figure><p><strong>3. Add payout details.<br> </strong>Upload a CSV, enter emails and amounts manually, or send the data through the API.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o08x8nBXKD3Gd19nae8AOA.png" /></figure><p><strong>4. Confirm and send.<br> </strong>The system checks the whitelisted IP and requests 2FA confirmation when enabled.</p><p><strong>5. Deliver SNSY.</strong></p><p>ChangeNOW PRO creates an account for new recipients and delivers funds in under one second with zero network fees.</p><p>Recipients can then hold, exchange, or withdraw their SNSY.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dVqI7_DnKDb_J7s_cncnfQ.png" /></figure><h3>Why Businesses Choose This Workflow</h3><p>Managing payouts is about more than delivering funds. Businesses also want to reduce manual work, control operating costs, and build processes that can scale as transaction volumes grow.</p><ul><li><strong>Lower operational costs.</strong> Benefit from a 0% payout service fee and zero network fees for ChangeNOW PRO payouts within the ecosystem.</li><li><strong>Less manual work.</strong> Sending payouts by email removes the need to collect, store, and update wallet addresses.</li><li><strong>Faster recipient experience.</strong> Recipients receive SNSY in under one second without waiting for on-chain transfers to reach their wallet.</li><li><strong>Enterprise automation.</strong> Automate recurring payouts through the dashboard or API, with infrastructure designed for reliable high-volume processing.</li><li><strong>One business ecosystem.</strong> Manage payments, payouts, conversions, and treasury workflows through a single crypto infrastructure instead of multiple providers.</li></ul><p>As crypto operations expand, payouts become just one part of a broader financial workflow. Bringing these processes together helps businesses scale more efficiently while reducing operational complexity.</p><h3>Conclusion</h3><p>Sensay has become a practical token for businesses that need to send recurring crypto payouts thanks to its multi-chain availability, growing ecosystem of integrations, and clear utility within the Sensay AI platform. As payout volumes grow, however, the token is only part of the equation. Operational efficiency depends on the infrastructure that helps businesses automate payments, manage treasury, and simplify recurring distributions.</p><p>As crypto becomes part of everyday business operations, choosing the right asset is only the first step. By combining Sensay with operational infrastructure like NOWPayments and ChangeNOW PRO, businesses can spend less time managing transactions and more time scaling their operations, strengthening partnerships, and rewarding the people who help them grow.</p><h3>Ready to Simplify SNSY Mass Payouts?</h3><p>If you’re looking for a faster way to distribute Sensay at scale, explore <a href="https://nowpaymentsinteractiveguide.pages.dev/">NOWPayments Zero-Fee Ecosystem Payouts</a>. Businesses can send SNSY by email, automate recurring payouts through the dashboard or API, and deliver funds to ChangeNOW PRO users in under one second with zero network fees.</p><p>→ <a href="https://zerofeepayouts.netlify.app/">Learn more about Zero-Fee Ecosystem Payouts</a></p><h3>FAQ</h3><p><strong>What is Sensay (SNSY)?</strong></p><p>Sensay is an AI ecosystem that enables the creation of lifelike digital replicas and AI-powered offboarding and knowledge management applications. SNSY is its native cryptocurrency, used for payments inside the Sensay platform, staking, partner rewards, and transferring value across Ethereum, Base, and Arbitrum networks.</p><p><strong>Why do businesses use Sensay (SNSY) for mass payouts?</strong></p><p>Businesses choose SNSY because it combines multi-chain accessibility with growing adoption by payment gateways, exchanges, and travel platforms. These qualities make it well suited for recurring payouts such as affiliate commissions, payroll, marketplace settlements, creator rewards, and gaming prizes.</p><p><strong>How do I automate Sensay (SNSY) mass payouts with NOWPayments?</strong></p><p>You can automate SNSY payouts through the NOWPayments Mass Payouts dashboard or API. Businesses can upload recipient data using a CSV file, enter it manually, or integrate payouts directly into their products through the API.</p><p><strong>Can I send Sensay (SNSY) to an email address with ChangeNOW PRO?</strong></p><p>Yes. Instead of collecting ERC-20 wallet addresses, businesses can send SNSY using a recipient’s email address. If the recipient already has a ChangeNOW PRO account, the funds are delivered immediately. If not, an account is created automatically during the payout process.</p><p><strong>What happens if the recipient doesn’t have a ChangeNOW PRO account?</strong></p><p>The payout still goes through. ChangeNOW PRO automatically creates an account for the recipient, who can then access the SNSY, exchange it for another supported cryptocurrency, or withdraw it to a personal wallet whenever they choose.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c497c5a54bf6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Managing Your SNSY Tokens Directly On-Chain]]></title>
            <link>https://asksensay.medium.com/managing-your-snsy-tokens-directly-on-chain-e0f1c0da78d8?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/e0f1c0da78d8</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Wed, 29 Jul 2026 09:43:28 GMT</pubDate>
            <atom:updated>2026-07-31T11:21:31.552Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Published: Wednesday, 29 July 2026 | Sensay</strong></p><h3>What is happening</h3><p>The hosted portal that many of you have used to claim vested tokens and to stake SNSY is being retired. Decubate, the provider that ran our vesting, staking, and liquidity mining front end (the TMS, ATV, and ATS services), is winding down its hosted portals, and they will stop working on <strong>1 August 2026</strong>.</p><p>We want to be completely clear about the most important part first. Your tokens are safe. Nothing is being deleted, moved, or locked. Every vesting schedule and every staking position lives inside smart contracts on the Ethereum blockchain, and those contracts continue to run exactly as before. What is going away is only the website that sat on top of them.</p><p>This means you can still claim, stake, unstake, compound, and withdraw everything yourself. From 1 August you will do it directly through a block explorer such as Etherscan rather than through the portal. This guide walks you through exactly how, using the real function names on our contracts.</p><p>We also know that a web interface is far friendlier than raw contract calls, so we are already working on a new interface for SNSY and will bring it to you as soon as it is ready. In the meantime, the steps below give you full, permanent control of your own tokens.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CbExuQB1wNXHWpRE0ZTdRw.png" /></figure><h3>The official contract addresses</h3><p>All of our contracts are on <strong>Ethereum mainnet</strong>. These are the addresses you will use.</p><ul><li>SNSY token: 0x82a605D6D9114F4Ad6D5Ee461027477EeED31E34</li><li>Vesting: 0x269A3b7b906FD423E70b14371126c10a37E2256c</li><li>Staking: 0x382c70620e42c2EF2b303b97bad1d9439Bf48ef9</li><li>Liquidity mining: 0xD9744Bd49248D78b424Ac1052145017E5E821c10</li><li>SNSY/WETH LP token (Uniswap v2): 0x778dcAA27e9c669A71c462b724a4e97Fc4F1Bcc5</li><li>WETH (used by liquidity mining): 0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2</li><li>Staking vault: 0xD97AD808a3B0c27c3F3966Cca198c066b3D26E8d</li><li>Farm: 0xE943c9d825bC45F92a92957eD86e5F5b99f78223</li></ul><p>Please only ever use addresses published through our official channels, and always check them character by character. If anyone sends you a different address in a direct message, treat it as a scam.</p><p>A quick note on two of these. The <strong>staking</strong> and <strong>liquidity mining</strong> contracts are proxy contracts, which simply means that when you open them on Etherscan you should use the <strong>Read as Proxy</strong> and <strong>Write as Proxy</strong> tabs. The SNSY token, the vesting contract, and the LP token are ordinary contracts, so for those you use the normal <strong>Read Contract</strong> and <strong>Write Contract</strong> tabs. More on this below.</p><h3>What you will need</h3><p>Before you start, make sure you have three things ready.</p><p>First, a Web3 wallet such as MetaMask, connected to Ethereum mainnet. This should be the same wallet address that holds your allocation or your staked position, because the contracts recognise you by your wallet address.</p><p>Second, a small amount of ETH in that wallet to pay for gas. Every on-chain action costs a network fee paid in ETH, so keep a little on hand.</p><p>Third, the correct contract address for the action you want to take, from the list above.</p><p>You will never need to enter your seed phrase or private key anywhere on a block explorer. If any page asks for it, leave immediately.</p><h3>The one concept that matters most: amounts and decimals</h3><p>This is the part that trips people up, so it is worth reading slowly.</p><p>When you use a website, you type “250” and it means 250 tokens. When you interact with a contract directly, the contract does not use decimal points. It works in the smallest possible unit, and SNSY, WETH, and the LP token all use 18 decimals. That means one whole token is represented as a 1 followed by eighteen zeros.</p><p>So when a function asks you for an amount, you enter the value in these smallest units:</p><ul><li>1 token is entered as 1000000000000000000</li><li>250 tokens is entered as 250000000000000000000</li><li>1,000 tokens is entered as 1000000000000000000000</li></ul><p>The rule is simply to take your whole-token amount and add eighteen zeros after it.</p><p>Etherscan often shows a small helper next to amount fields that lets you type the whole number and select the “10¹⁸” option to do this conversion for you. If you see it, you can use it. If you are ever unsure, enter the full value manually as shown above, because getting the number of zeros wrong is the most common mistake people make.</p><h3>How to reach a contract on Etherscan</h3><p>The following steps are the same for all of our contracts. The only differences are the address you use, whether you use the standard or the “as Proxy” tab, and the specific function you call.</p><ol><li>Go to <a href="https://etherscan.io./">https://etherscan.io.</a></li><li>Paste the contract address into the search bar and open the contract page.</li><li>Click the <strong>Contract</strong> tab. A verified contract shows a green tick here, and all of ours are verified.</li><li>To view information, use <strong>Read Contract</strong> (or <strong>Read as Proxy</strong> for the staking and liquidity mining contracts). To take an action, use <strong>Write Contract</strong> (or <strong>Write as Proxy</strong>).</li><li>To take any action, click <strong>Connect to Web3</strong> and approve the connection in your wallet. Make sure the connected address is the one that holds your position.</li></ol><p>Once connected, you choose the function you want, enter any details it asks for, click <strong>Write</strong>, and confirm the transaction in your wallet. The sections below tell you exactly which functions to use in each case.</p><h3>Guide 1: Staking SNSY</h3><p>Staking works slightly differently from vesting, and it is worth understanding why before you start, because it explains something that confuses people when they first look on-chain.</p><p>When you staked through the portal, your tokens did not go into the staking contract under your own wallet. They went into an auto-compounding vault, which stakes into the staking contract on behalf of everyone at once, as a single pooled position. The vault keeps track of what each person is owed using <strong>shares</strong>.</p><p>This is why, if you look up your own address on the staking contract, it appears to hold nothing. Your position is real and fully intact, but it is recorded in the vault rather than in the staking contract. The vault is where you check it, and the vault is where you withdraw from.</p><p>The vault address is:</p><p>0xD97AD808a3B0c27c3F3966Cca198c066b3D26E8d</p><p>You do not have to take our word for the fact that these two contracts belong together. Two quick reads on Etherscan confirm it, and we would encourage you to do them:</p><ul><li>On the <strong>staking contract</strong>, call compounderContract(). It returns the vault address above.</li><li>On the <strong>vault</strong>, call masterchef(). It returns the staking contract address.</li></ul><p>There is nothing you need to do to migrate, move, or re-register your position. It sits in the vault exactly as it did before, and it will stay there indefinitely. If anyone ever tells you to move or re-stake your tokens before a deadline, that is a scam and you should ignore it.</p><h3>Checking your position</h3><p>Open the vault on Etherscan and go to the <strong>Read as Proxy</strong> tab. You will need your pool ID, which is the same pool number you originally staked into, and your own wallet address.</p><p><strong>users</strong> takes a pool ID and your address, and returns four values: your shares, the date you deposited, the amount of SNSY you originally staked, and how much you have claimed.</p><p><strong>getRewardOfUser</strong> takes your address and a pool ID, in that order, and shows what your position has earned so far.</p><p><strong>getPricePerFullShare</strong> takes a pool ID and tells you what one share is currently worth. Multiply your shares by this figure and divide by 1e18 to see what your position is worth in SNSY today. This number is above 1 for every pool, because rewards have already accrued into it.</p><p><strong>canUnstake</strong> takes your address and a pool ID, and returns true if your lock period has passed and you are free to withdraw.</p><p>Two things are worth knowing here. Shares are not the same number as SNSY, so do not be alarmed if your share balance looks different from the amount you staked. And rewards compound automatically inside the vault, so there is nothing separate to claim along the way. When you withdraw, you receive your stake and everything it has earned together, in one transaction.</p><h3>Withdrawing</h3><p>Once canUnstake returns true, go to the vault&#39;s <strong>Write as Proxy</strong> tab and connect your wallet.</p><p><strong>withdrawAll</strong> takes just a pool ID and withdraws your entire position for that pool. This is the simplest option and the one most people will want.</p><p><strong>withdraw</strong> takes a pool ID and an amount, and withdraws part of your position. The important detail is that the amount is in <strong>shares, not SNSY</strong>. Read your share balance from users first and use a figure no larger than that. If you enter your SNSY amount by mistake the transaction will simply fail rather than doing anything unexpected, but it is easier to get it right first time.</p><p>Early withdrawal is switched off on-chain for every pool, so a position that is still within its lock genuinely cannot be taken out ahead of its unlock date, by us or by anyone. Once the lock has passed there is no deadline at all, and you can withdraw whenever suits you.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HdAhTDcKaWU3SzqbM8Laag.png" /></figure><h3>The earlier farming contract</h3><p>A small number of wallets used an earlier SNSY farming contract, separate from the staking pools above. It was not mentioned in the first version of this guide, and we are grateful to the community member who pointed that out.</p><p>The address is:</p><p>0xE943c9d825bC45F92a92957eD86e5F5b99f78223</p><p>This one has a single pool, pool 0, with a 14-day lock, and its reward period ended in May 2024. If you used it, your position is unaffected and remains fully withdrawable.</p><p>Unlike the staking pools, this contract holds positions under your own wallet directly, so there is no vault involved. On the <strong>Read as Proxy</strong> tab, users with pool ID 0 and your address shows your stake, and payout with the same inputs shows any rewards still owed. On the <strong>Write as Proxy</strong> tab, claim with pool ID 0 pays out your rewards, and unStake with pool ID 0 and an amount returns your stake. Withdrawing also pays out any outstanding rewards at the same time.</p><p>If you are unsure whether you ever used this contract, look up your own address against it using users as above. A result of zero means you did not.</p><h3>Guide 2: Liquidity mining (Uniswap v2)</h3><p>The liquidity mining contract has a single pool, <strong>pool ID 0</strong>. It rewards you in SNSY for providing liquidity to the SNSY/WETH pool on Uniswap v2, and it has a 30-day lock. Unlike a simple staking contract, this one manages the Uniswap liquidity for you, so it works with SNSY and WETH (wrapped ETH) rather than with a plain deposit.</p><p>For most people the two actions that matter now are checking and claiming rewards, and withdrawing after the lock. Both are straightforward.</p><p><strong>Check your position and rewards.</strong> Open the liquidity mining contract and go to <strong>Read as Proxy</strong>. Use payout with pool ID 0 and your wallet address to see your claimable SNSY rewards, and users with the same inputs to see your staked position.</p><p><strong>Claim rewards.</strong> On <strong>Write as Proxy</strong>, use claim with pool ID 0, or claimAll, then confirm in your wallet.</p><p><strong>Withdraw and remove your liquidity.</strong> Once your 30-day lock has passed, use unlockAndRemoveLP on <strong>Write as Proxy</strong>. It takes the pool ID (0), the amount of your staked liquidity to remove in smallest units, and two minimum amounts, one for SNSY and one for WETH, which protect you against price slippage while your liquidity is being removed. Setting both minimums to 0 will always succeed and is the simplest option, though it offers no slippage protection, so during volatile periods it is safer to set them a little below the amounts you expect to receive. Your SNSY and WETH are returned to your wallet.</p><p><strong>Adding new liquidity (advanced).</strong> It is possible to add and lock new liquidity directly using the addLiquidityAndLock function, but this is the most involved action on any of our contracts. It requires approving both SNSY and WETH to the mining contract and setting sensible minimum amounts, and mistakes can cause the transaction to fail. If you want to add new liquidity, we would gently suggest waiting for the new interface, or reaching out to the team first so we can guide you through it.</p><h3>Guide 3: Claiming vested tokens</h3><p>The vesting contract organises allocations into numbered groups, which it calls <strong>options</strong>. To claim, you need to know your option number. Public sale participants are generally in option <strong>3</strong> (named “Public Sale”) or option <strong>7</strong> (named “Public Sale Enj”). If you received tokens through any other route, the team can tell you your exact option number, so please just ask through our official channels.</p><p><strong>Step 1: Find and confirm your option.</strong> Open the vesting contract and go to <strong>Read Contract</strong>. Find the function getReleasableAmount, which takes two inputs, an option number and a wallet address. Enter your likely option number and your wallet address, then run it. If it returns a number greater than zero, that is the amount (in smallest units) you can currently claim, and you have the right option. If it returns zero and you believe you have tokens due, try your other likely option or confirm the number with the team. The related read function getVestAmount shows your total vested amount for an option, and getWhitelist shows your original allocation.</p><p><strong>Step 2: Claim.</strong> Go to <strong>Write Contract</strong>, connect your wallet, and find the function claimDistribution. It takes two inputs, the option number and the wallet address. Enter your option number and your own wallet address, click <strong>Write</strong>, and confirm in your wallet. Your unlocked tokens will arrive in that wallet once the transaction confirms.</p><p>You can repeat this whenever more of your allocation unlocks over time.</p><h3>If a contract shows no read or write functions</h3><p>If a contract page does not show the usual Read or Write tabs, or shows very little, it is because that contract is a proxy, where the logic lives at a separate address. This is the case for our staking and liquidity mining contracts. Simply use the <strong>Read as Proxy</strong> and <strong>Write as Proxy</strong> tabs, which Etherscan provides for exactly this reason, and everything will appear as expected.</p><h3>A short safety checklist</h3><p>Before you confirm any transaction, run through this quickly.</p><ul><li>You are on the genuine etherscan.io, and the contract shows a green verified tick.</li><li>The contract address matches the official one above, checked character by character.</li><li>Your connected wallet is the address that holds your position.</li><li>Amounts include the correct eighteen zeros.</li><li>You have never been asked for your seed phrase, because a legitimate process never asks for it.</li></ul><h3>A new interface is on the way</h3><p>We know that claiming and staking through a block explorer is more manual than the portal you were used to. This guide exists so that you always retain full control of your own tokens, with no dependency on anyone else’s website staying online. That is genuinely a healthier position to be in.</p><p>At the same time, we are building a new, friendly interface for SNSY vesting, staking, and liquidity mining, and we will release it as soon as it is ready. When it launches we will announce it through our official channels. Until then, everything you need is on-chain and in your hands.</p><h3>Getting help</h3><p>If you are unsure which contract, option, or pool applies to you, reach out through our official community channels and we will point you to the right place. Please be cautious of anyone offering to “help” you in private messages or asking you to visit an unfamiliar site, as this is a common tactic used by scammers during transitions like this one.</p><p>Your tokens are safe, and you remain fully in control.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e0f1c0da78d8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Knowledge Ledger: How Web3 Can Rewire Enterprise Knowledge Management]]></title>
            <link>https://asksensay.medium.com/the-knowledge-ledger-how-web3-can-rewire-enterprise-knowledge-management-bb236c56e473?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/bb236c56e473</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Thu, 16 Jul 2026 21:43:19 GMT</pubDate>
            <atom:updated>2026-07-16T21:43:19.628Z</atom:updated>
            <content:encoded><![CDATA[<p>Every enterprise runs on knowledge it cannot see. Not the documents in the wiki or the tickets in Jira, but the tacit layer underneath: why a decision was made, who to call when a system breaks, the workaround that never got written down. This knowledge lives in people’s heads, and it walks out the door when they do. The average knowledge worker spends a fifth of their week just looking for information that already exists somewhere in the organization. When someone leaves, a slice of institutional memory leaves with them, unrecoverable.</p><p>Knowledge management has spent thirty years trying to solve this with better databases, better search, and lately better AI. Those tools help. But they leave three structural problems untouched: <strong>ownership</strong>, <strong>provenance</strong>, and <strong>incentives</strong>. Web3 is interesting precisely because it attacks those three, not because it adds another place to store files.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*Fajzc2yIcXIlQ-6KZhb1jQ.png" /></figure><h3>The three cracks in the foundation</h3><p>Consider how knowledge actually moves through a company today.</p><p><strong>Ownership sits with the platform, not the organization.</strong> Your knowledge base lives inside a SaaS vendor’s walled garden. Export is painful, portability is a fiction, and the moment you switch tools, your carefully structured knowledge graph fractures into a pile of markdown. The enterprise generated the knowledge but doesn’t truly control the substrate it lives on.</p><p><strong>Provenance is an afterthought.</strong> Who wrote this? When? Was it reviewed? Is it still true? Most systems answer these questions poorly, if at all. In a world now flooded with AI-generated content, the inability to verify where a claim came from is no longer a minor annoyance, it’s a liability. When an LLM confidently cites an internal document, can anyone trace that document back to a trustworthy human source?</p><p><strong>Incentives are misaligned to the point of comedy.</strong> We ask employees to document their knowledge for the good of the company, then measure and reward them on everything except that. Contributing knowledge is unpaid, invisible labor that mostly benefits your successor. So people don’t do it, and the wiki rots.</p><p>These aren’t tooling problems. They’re architecture problems. And architecture is what web3 is actually about.</p><h3>What the primitives actually offer</h3><p>Strip away the speculative noise and web3 is a small set of primitives: cryptographic identity, shared immutable ledgers, programmable incentives, and content-addressed storage. Mapped onto knowledge management, they do real work.</p><p><strong>Verifiable provenance through decentralized identity.</strong> When a contribution is signed with a decentralized identifier (DID) and a verifiable credential, you get a tamper-evident record of who said what and on what authority. A troubleshooting guide isn’t just text in a database; it’s an attested artifact linked to the engineer who wrote it and the reviewers who validated it. Years later, when that engineer is gone, the provenance survives. In an AI-saturated environment, this is the difference between a knowledge base you can trust and one you merely hope is accurate.</p><p><strong>Durable, portable knowledge assets.</strong> Content-addressed storage (like IPFS) references knowledge by its cryptographic hash rather than by a location inside a vendor’s system. The artifact is identified by <em>what it is</em>, not <em>where it lives</em>. That breaks the vendor lock-in at the storage layer and makes knowledge genuinely portable across tools, teams, and even organizational boundaries. Your knowledge outlives your software contracts.</p><p><strong>Programmable incentives that finally reward contribution.</strong> This is the primitive most likely to change behavior. Token mechanics let an organization reward knowledge contribution directly, transparently, and in proportion to value. Imagine a system where documenting a critical process, answering a colleague’s question, or capturing a departing expert’s playbook earns verifiable, on-chain recognition tied to real internal value. The economics of “why should I bother writing this down” invert. Contribution becomes a credited act rather than a thankless one.</p><p><strong>Shared governance across boundaries.</strong> DAO structures allow multiple parties, business units, supply-chain partners, even competitors in a consortium, to co-own and govern a knowledge commons without any single party controlling it. For industry standards bodies, research collaborations, and multi-vendor ecosystems, a neutral, jointly-governed knowledge ledger solves a trust problem that no centralized platform can.</p><h3>The killer application: knowledge transfer at the edges</h3><p>The sharpest use case sits at the moments of highest knowledge risk: <strong>offboarding, succession, and cross-organizational handoff.</strong></p><p>When a senior employee retires after twenty years, the enterprise is facing an irreversible loss. Capture their tacit knowledge through structured, AI-guided interviews, attest each captured insight with the departing expert’s verifiable signature, store it in a portable, content-addressed format, and reward the act of transfer through a transparent incentive layer, and you’ve built something no traditional KM system can: a trustworthy, durable, provenance-rich record of institutional memory that survives the person, the team, and even the tools that captured it.</p><p>The same architecture applies to mergers, where two companies must reconcile incompatible knowledge systems, and to supply chains, where partners need to share operational knowledge without surrendering it to each other’s platforms.</p><h3>The honest caveats</h3><p>Web3 evangelists oversell, so let me be blunt about the limits. <strong>Most of your knowledge does not belong on a blockchain.</strong> Ledgers are slow, expensive, and public by default, which is exactly wrong for high-volume, private, frequently-edited content. The right pattern is almost always hybrid: keep the bulk of content in fast conventional storage, and put only the <em>provenance, attestations, and incentive records</em> on-chain, the thin layer of trust and reward that conventional systems handle worst.</p><p>The user experience also has to disappear. No employee will manage a wallet to file a document. The cryptography must be invisible, abstracted entirely behind the same clean interfaces people already use in Slack, Teams, and their browser. And the regulatory questions around data residency, the right to be forgotten, and tokenized incentives are real and unresolved. Immutability and GDPR’s erasure requirements are in genuine tension.</p><h3>The shift worth making</h3><p>The point of web3 for knowledge management is not decentralization for its own sake. It’s a reframing: <strong>treat organizational knowledge as an owned, provenanced, and incentivized asset rather than as disposable content locked inside rented software.</strong></p><p>Enterprises that make that shift get knowledge bases they can trust in an age of AI slop, that outlast any single vendor, and that people are actually motivated to feed. The database was the last thirty years of knowledge management. The ledger, used surgically and invisibly, may define the next thirty.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bb236c56e473" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Will You Survive Your Own Death? Ten Books That Take the Question Seriously]]></title>
            <link>https://asksensay.medium.com/will-you-survive-your-own-death-ten-books-that-take-the-question-seriously-ec7afac16085?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/ec7afac16085</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Thu, 04 Jun 2026 18:18:56 GMT</pubDate>
            <atom:updated>2026-06-04T18:18:56.702Z</atom:updated>
            <content:encoded><![CDATA[<h3>A reading list on mind uploading, consciousness, and digital immortality from the last three years, when the field stopped being science fiction</h3><p>In June 2023, the neuroscientist Christof Koch lost a bet he’d been carrying for twenty-five years. He had wagered David Chalmers a case of fine wine that we’d find the neural footprint of consciousness within a quarter-century. We didn’t. Koch paid up with a case of Madeira, and in doing so handed the field a small, honest monument to how hard this problem remains.</p><p>I find that story clarifying. The people closest to the question of what consciousness <em>is</em> — and therefore whether it can be copied, preserved, transferred, or rebuilt in silicon — are not the ones selling certainty. They’re the ones losing bets gracefully and going back to work.</p><p>That’s the spirit I want for this list. If you care about mind uploading, digital immortality, or the harder question underneath both — whether <em>you</em> could survive the death of your biological brain — the last three years have been the richest publishing window the field has ever seen. Three conversations that used to run in separate rooms finally collided: the philosophy of personal identity, the neuroscience of consciousness, and the applied ethics of AI. Almost everything worth reading sits at one of those collision points.</p><p>Here are ten works — books, papers, and one genuinely field-shaking experiment — that earn their place. I’ve ordered them roughly from most accessible to most technical, and I’ll tell you not just what each one says but why it matters and who should actually read it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QZZZhBVJARFTs4gnZ_hQiQ.png" /></figure><h3>The books that frame the whole problem</h3><h3>1. The Future Loves You: How and Why We Should Abolish Death — Ariel Zeleznikow-Johnston (2024)</h3><p>If you read only one book on this list, read this one. Zeleznikow-Johnston is a neuroscientist at Monash University, and his argument is the most decision-relevant thing published in years: death should be redefined not as the moment your heart stops, but as the irreversible loss of your brain’s structural connectome — the wiring diagram that encodes who you are.</p><p>The implication is radical and surprisingly well-argued. If that structure can be preserved — through aldehyde-stabilized cryopreservation and related techniques — then “death” as we currently define it may be a premature verdict. The book braids serious neuroscience with a Derek Parfit–style argument that what matters for your survival is psychological continuity, not the particular meat your mind currently runs on.</p><p>It carries endorsements from George Church and Ken Hayworth, two of the most credible names in brain preservation, and landed on <em>The Telegraph</em>’s best science books of 2024. Read it first, because if its case lands for you, every other book here reads differently — uploading stops being a thought experiment and becomes an operational question.</p><p><strong>Best for:</strong> anyone weighing brain preservation as a real option, not a punchline.</p><h3>2. Then I Am Myself the World: What Consciousness Is and How to Expand It — Christof Koch (2024)</h3><p>This is the other side of the coin, written by the man who lost the bet. Koch’s book is the most readable defense of Integrated Information Theory (IIT) — the framework that a now-infamous 2023 open letter branded “pseudoscience,” signed by 124 researchers. Koch doesn’t dodge the fight; he leans into it.</p><p>What makes the book matter for <em>our</em> question is one quietly devastating chapter. Koch argues that digital computers, however capable, cannot host human-level consciousness, because each transistor connects to only a handful of others — orders of magnitude below the thousands of synaptic partners every cortical neuron maintains. Consciousness, in his telling, depends on the <em>causal structure</em> of the substrate, and silicon simply doesn’t have enough of the right kind. Ned Block, reviewing it in <em>Science</em>, called it charming and illuminating.</p><p>If Koch is right, the dream of uploading yourself to the cloud is dead on arrival, and the only credible path runs through biology. Read him directly against Anil Seth (below) for the sharpest disagreement in the field.</p><p><strong>Best for:</strong> the strongest popular case that consciousness is biological, measurable, and not portable to a server.</p><h3>3. The Afterlife of Data — Carl Öhman (2024)</h3><p>Here’s the half of digital immortality nobody romanticizes: your data outlives you whether you plan for it or not. Öhman, a political scientist at Uppsala, argues that your digital remains are conceptually analogous to a corpse — an “informational body” that carries something of your identity and that someone, somewhere, will end up stewarding.</p><p>The scale is staggering. Öhman’s earlier modeling suggests the dead could outnumber the living on platforms like Facebook within decades, climbing toward billions of deceased profiles by the end of the century. He dissects the emerging “digital afterlife industry” — including the griefbots now being built to simulate conversations with the dead — and argues that this informational inheritance is too important to leave entirely to private platforms.</p><p>It made <em>The Economist</em>’s and <em>Nature</em>’s best-of-2024 lists for good reason. This is the book to read about <em>legacy</em> specifically — what survives of you by default, and who controls it.</p><p><strong>Best for:</strong> thinking concretely about your own digital legacy and the governance that will decide its fate.</p><h3>4. The Weirdness of the World — Eric Schwitzgebel (2024)</h3><p>A palate cleanser, and a necessary one. Schwitzgebel, a sharp analytic philosopher of mind, argues for what he calls “Universal Bizarreness”: every coherent theory of how mind relates to the cosmos ends up violating common sense somewhere. So we should hold our metaphysics loosely and take the wild options seriously.</p><p>His chapters ask whether the United States might literally be conscious as a collective entity, whether we live in a simulation, and whether consciousness is even intelligible in principle. The <em>Times Literary Supplement</em> praised the way his intelligence and sense of wonder produce humbling conclusions.</p><p>The payoff for this list: uploading, machine consciousness, panpsychism, and simulation theory are all, in their own way, <em>less</em> absurd than smugly refusing to consider them. It’s a book that teaches intellectual humility without descending into mush.</p><p><strong>Best for:</strong> the philosophically curious reader who wants permission to take strange ideas seriously.</p><h3>The frontier of machine minds</h3><h3>5. The Edge of Sentience — Jonathan Birch (2024)</h3><p>Birch chaired the UK government review that extended legal recognition of sentience to octopuses, crabs, and lobsters. Here he builds a general framework for acting ethically toward any system whose inner life is uncertain — fetuses, brain organoids, patients with disorders of consciousness, and AI.</p><p>His key move is the idea of a “sentience candidate”: any system where the probability of sentience is credible and non-negligible deserves proportionate precaution, even without proof. It won the Royal Institute of Philosophy’s 2025 transdisciplinary prize, and it’s openly available. For anyone making real decisions about AI systems, it’s the most defensible policy framework going — a way to act responsibly under deep uncertainty rather than waiting for a certainty that may never come.</p><p><strong>Best for:</strong> anyone who needs a rigorous basis for moral decisions about minds we can’t yet verify.</p><h3>6. “Conscious Artificial Intelligence and Biological Naturalism” — Anil Seth (2025)</h3><p>Seth wrote the bestselling <em>Being You</em>; this <em>Behavioral and Brain Sciences</em> target article is his most rigorous statement yet on whether machines can be conscious. His answer is a careful no — at least not along current lines. Consciousness, he argues, is rooted in our nature as <em>living organisms</em>, not in computation as such. Real artificial consciousness only becomes plausible as AI grows more brain-like or life-like, which today’s systems are not.</p><p>His companion essay won the 2025 Berggruen Prize essay competition out of roughly 3,000 submissions worldwide. As a BBS target article, it’ll arrive surrounded by open peer commentaries — the field’s gold standard for structured disagreement.</p><p><strong>Best for:</strong> the strongest current argument that scaling language models will not, by itself, wake anything up.</p><h3>7. “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” — Butlin, Long, et al. (2023)</h3><p>This is the paper that turned “do AIs feel?” into something an engineer can actually work with. Its nineteen authors — including David Chalmers and Yoshua Bengio — derived concrete <em>indicator properties</em> of consciousness from leading neuroscientific theories, then assessed real AI systems against them. The verdict: nothing today is conscious, but no obvious technical barrier rules out building something that would tick several boxes.</p><p>This report quietly catalyzed the AI-welfare research now happening inside major labs. Its follow-up, “Taking AI Welfare Seriously,” was co-authored by the researcher Anthropic later hired to run exactly that program. Whatever you think of the conclusions, this is the document that made the question respectable.</p><p><strong>Best for:</strong> the engineer’s version of the consciousness question — falsifiable criteria instead of vibes.</p><h3>8. “The Error Theory of LLM Consciousness” / “Chatbot Epistemology” — Susan Schneider (2025)</h3><p>Schneider is fascinating because she changed her mind in public. She once championed behavioral tests for machine consciousness; now she argues the opposite. Her “error theory” holds that when a language model reports having feelings, that’s an artifact of training on a human corpus saturated with descriptions of consciousness — a “crowdsourced neocortex” mimicking inner life it doesn’t have.</p><p>Her companion piece extends the worry to us: the slow erosion of our own cognition and identity as we offload more thinking to AI companions, a “boiling frog” problem we won’t notice until it’s done. The work has already drawn multiple published replies — a sign it hit a nerve.</p><p><strong>Best for:</strong> the most rigorous skeptic’s case against LLM consciousness, from someone who used to believe the opposite.</p><h3>The experiment and the blueprint</h3><h3>9. “Adversarial Testing of Global Neuronal Workspace and Integrated Information Theories” — Cogitate Consortium, Nature (2025)</h3><p>The single most important empirical paper of this window. Rival camps — proponents of IIT and of Global Neuronal Workspace Theory — agreed <em>in advance</em> on what each theory predicted and how results would be interpreted, then ran the experiment together. This is how science is supposed to work and almost never does.</p><p>The results bruised both theories. IIT’s predicted sustained synchronization in the back of the brain didn’t materialize; the workspace theory struggled with the lack of expected activity in the prefrontal cortex. No clean winner. It gave the 2023 “pseudoscience” letter real empirical teeth while showing that the leading alternative isn’t unscathed either. The honest takeaway: nobody currently has a complete theory of consciousness, and the people who tell you otherwise are selling something.</p><p><strong>Best for:</strong> seeing what testing these grand theories actually looks like — and why humility is warranted.</p><h3>10. “Griefbots, Deadbots, Postmortem Avatars” — Hollanek &amp; Nowaczyk-Basińska (2024)</h3><p>We end where many of us will actually engage with this: the bots. Researchers at Cambridge’s Leverhulme Centre wrote the most-cited recent treatment of the “digital afterlife industry” — the companies building chatbots that simulate the dead.</p><p>Their framework distinguishes the data donor (the deceased), the recipient (whoever holds the data), and the interactant (whoever talks to the simulation), and from that derives concrete ethics: dignified “retirement” protocols for deadbots, bans on advertising through simulated loved ones, and a proposed “Digital Do Not Reanimate” directive. As products in this space move from concept to market, this is the closest thing we have to a responsible-use blueprint.</p><p><strong>Best for:</strong> anyone considering building — or becoming — a legacy AI.</p><h3>11. BONUS: “Endless: Life After Upload” — Thomson (2026)</h3><p>The technology to upload human consciousness is no longer science fiction — it’s an engineering problem. Within decades, the first successful mind uploads may occur. And then everything changes.</p><p>In <em>Endless</em>, Dan Thomson ventures beyond the theoretical into the lived reality of a post-upload world. What does society look like when your grandfather exists as a digital being? When couples face relationships measured in centuries? When humanity itself must be redefined?</p><p>This is not utopia. This is not dystopia. This is something far more interesting: the messy, complex, deeply human future that awaits when mortality becomes a choice.</p><h3>How to actually read these</h3><p>If you want this list to change how you think rather than just decorate a shelf, here’s the order I’d suggest.</p><p>Start with <strong>Zeleznikow-Johnston</strong>. Let him reframe death for you. Then read <strong>Öhman</strong> and the <strong>griefbots</strong> paper together as the <em>legacy</em> problem — what survives of you by default, with or without uploading. Next, set <strong>Koch against Seth</strong> deliberately, back to back: causal structure versus biological life as the secret ingredient. If they’re both right, cloud uploading is hopeless and biology is the only road. Use <strong>Butlin et al.</strong> and <strong>Cogitate</strong> as the empirical ground beneath all of it. Bring in <strong>Birch</strong> when you need to <em>act</em> under uncertainty, and keep <strong>Schwitzgebel</strong> and <strong>Schneider</strong> nearby as correctives — one insisting every option is weird, the other insisting our favorite intuition is wrong.</p><p>A few honest caveats. The IIT controversy is unresolved; Koch’s book is its best defense, and you should read it knowing serious people contest his central theory. Some of these papers are still working through formal publication, so their reception is live. And if you want long-form background, Seth’s <em>Being You</em> and Schneider’s <em>Artificial You</em> — both just before this window — are referenced by nearly everything above.</p><p>Finally, a filter worth keeping: the shelves and algorithms are full of “quantum consciousness” and frequency-raising titles that belong nowhere near a serious list. The bar here was peer review or substantive reviews in places like <em>Nature</em>, <em>Science</em>, and <em>The Economist</em>. The honest work in this field isn’t the work that promises you’ll live forever. It’s the work that takes the question seriously enough to admit how much we still don’t know — and goes back to the lab anyway.</p><p><em>What would change my mind on this list? A credible third-party validation of long-term brain-preservation fidelity would promote Zeleznikow-Johnston to required reading. A direct empirical test of Seth’s biological-naturalism claim against connectionist accounts would earn its own slot. And if Chalmers or Schneider publishes a full monograph on uploading and identity in 2026, it goes straight in.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ec7afac16085" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Immortality Tech, State of the Field — May 2026: What’s Actually Working, What’s Hype, and Where…]]></title>
            <link>https://asksensay.medium.com/immortality-tech-state-of-the-field-may-2026-whats-actually-working-what-s-hype-and-where-f2cd50dd331d?source=rss-96b57f9400dc------2</link>
            <guid isPermaLink="false">https://medium.com/p/f2cd50dd331d</guid>
            <dc:creator><![CDATA[Sensay]]></dc:creator>
            <pubDate>Thu, 14 May 2026 18:25:05 GMT</pubDate>
            <atom:updated>2026-05-14T18:25:05.904Z</atom:updated>
            <content:encoded><![CDATA[<h3>Immortality Tech, State of the Field — May 2026: What’s Actually Working, What’s Hype, and Where We’re Heading</h3><h3>TL;DR</h3><ul><li><strong>The biology side is finally translating.</strong> Partial epigenetic reprogramming reached the clinic for the first time in Q1 2026 (Life Biosciences’ ER-100), Loyal’s LOY-002 senior-dog longevity drug cleared two of three FDA technical sections by December 2025, and the Hevolution-led “healthspan” funding boom pushed 2024 sector capital to $7.33 billion — even as flagship senolytics company Unity Biotechnology liquidated in September 2025.</li><li><strong>The hardware-of-the-body side made bigger leaps than expected.</strong> Tim Andrews lived 271 days with an eGenesis pig kidney (a world record), the United Therapeutics EXPAND trial — the first regulated clinical xenotransplant program — performed its first transplant November 3, 2025, and three pig-liver transplants in China demonstrated the procedure is technically feasible. The MICrONS mouse-visual-cortex connectome (April 2025) and Neuralink’s expansion to 12 implanted patients with iPad/iPhone control mean the digital-immortality stack is no longer purely theoretical.</li><li><strong>But “immortality” remains a category error.</strong> The Longevity Escape Velocity Foundation’s RMR-1 mouse study — the most ambitious combination-therapy lifespan experiment ever run — produced only a ~4-month median extension and missed its 12-month “robust” target, and Aubrey de Grey still puts 50/50 odds on LEV at 2036, not soon. The honest read: we are bending the actuarial curve, not breaking it.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qqtPAe1zZ0o6B_09_1OLOg.jpeg" /></figure><h3>Key Findings</h3><ol><li><strong>Cellular reprogramming has crossed from preclinical to clinical</strong>, with Life Biosciences receiving FDA IND clearance for ER-100 (an OSK partial-reprogramming gene therapy for optic neuropathies) in January 2026, and Altos Labs hiring Joan Mannick as CMO in 2025 — the strongest tell yet that the $3 billion stealth giant is preparing trials.</li><li><strong>Senolytics had a brutal 2025.</strong> Unity Biotechnology’s lead asset UBX1325 missed its Phase 2b ASPIRE primary endpoint in March 2025; shareholders voted to liquidate the company in September 2025. The senescence-clearance thesis isn’t dead, but the “one senolytic, many indications” pitch is.</li><li><strong>The first regulated xenotransplant clinical trial is live.</strong> United Therapeutics’ UKidney EXPAND trial began November 3, 2025 at NYU Langone; eGenesis’s parallel multi-patient expanded access program at MGH set a world-record 271-day pig-kidney survival in Tim Andrews before failure in October 2025.</li><li><strong>Brain-computer interfaces went from one patient to a clinical fleet.</strong> Neuralink had 12 implanted users by September 2025 (cumulative 15,000 device-hours), Synchron raised a $200M Series D in November 2025 with a pivotal trial targeting 2026 FDA submission, and Precision Neuroscience holds 510(k) clearance for temporary cortical interfaces.</li><li><strong>Connectomics scaled by an order of magnitude.</strong> The MICrONS consortium published the first millimeter-cube mouse visual cortex connectome (~84,000 neurons, ~500 million synapses) in <em>Nature</em> in April 2025 — about 20× the complexity of the fully mapped fruit-fly brain (FlyWire, 2024).</li><li><strong>Rapamycin is the only “longevity drug” with real clinical data in healthy humans</strong> as of 2026: the AgelessRx PEARL trial (NCT04488601, 114 completers, published April 2025 in <em>Aging</em>) found 5–10 mg weekly rapamycin safe over 48 weeks with modest healthspan markers — but no proven lifespan effect. Metformin’s TAME trial remains stuck in funding limbo and has been partly absorbed into ARPA-H plus a parallel Eli Lilly GLP-1 trial.</li><li><strong>The “Bryan Johnson effect” has fully arrived.</strong> Johnson’s Blueprint protocol, the Don’t Die documentary on Netflix, and the launch of his Blueprint Biomarkers platform (a $365/year service running ~140 biomarkers on the same TruDiagnostic epigenetic clock he uses) have made DunedinPACE a household metric in the biohacker community — even as the American Council on Science and Health and others note his “n=1 case study” doesn’t validate the regimen.</li><li><strong>The aging-research framework has consolidated</strong> around López-Otín et al.’s 12 Hallmarks of Aging (Cell, January 2023), which added disabled macroautophagy, chronic inflammation, and dysbiosis to the original nine — now the de facto target list for every serious longevity biotech.</li><li><strong>Money is concentrated and bifurcated.</strong> Cellular reprogramming captured ~62% of the $7.5B raised in longevity 2022–2025 — Altos, Retro ($1B Series A in 2025 at a $5 billion valuation), NewLimit ($175M in 2025 including Eli Lilly’s first longevity check), Life Biosciences. Cryonics, by contrast, remains a $6–11M annual market with ~500 patients stored globally.</li></ol><h3>Details</h3><h3>1. Biological Longevity &amp; Rejuvenation</h3><p><strong>Partial reprogramming reaches the clinic.</strong> The single most consequential 2025–2026 event in geroscience is that Life Biosciences — the David Sinclair-co-founded biotech — received FDA IND clearance for <strong>ER-100</strong> in January 2026, with Phase 1 initiation in Q1 2026 for open-angle glaucoma and non-arteritic anterior ischemic optic neuropathy (NAION). ER-100 is an AAV-delivered intravitreal gene therapy expressing three Yamanaka factors — Oct4, Sox2, Klf4 (OSK) — under doxycycline control, building directly on the 2020 <em>Nature</em> paper from David Sinclair’s lab showing OSK can restore vision in aged mice. CSO Sharon Rosenzweig-Lipson called it “the first ever cellular rejuvenation therapy using partial epigenetic reprogramming to reach human clinical trials.” This is the first time a Yamanaka-factor therapy has been put into a human body.</p><p><strong>Altos Labs</strong>, the $3 billion stealth giant funded by Bezos, Yuri Milner, and Saudi-linked investors, sent the strongest “we’re going clinical” signal of 2025 with the hire of <strong>Joan Mannick</strong> — the rapamycin-aging expert formerly of resTORbio and Tornado Therapeutics — as Chief Medical Officer and head of product development. Altos’s founding scientist Juan Carlos Izpisua Belmonte published in <em>Cell</em> in 2025 on “mesenchymal drift” as a unifying aging mechanism, and the company has been testing reprogramming therapies on <strong>ex vivo perfused organs</strong> (rejuvenating kidneys outside the body for transplant) — a clever regulatory end-run that avoids the in-vivo cancer risk that dogged early reprogramming work.</p><p><strong>Retro Biosciences</strong>, the Sam Altman-backed startup ($180M seed in 2022 from Altman personally), dosed its first human patient in 2025 with <strong>RTR242</strong>, a small-molecule lysosomal-acidification autophagy enhancer aimed at neurodegeneration. Per the Financial Times (January 24, 2025) and STAT’s Allison DeAngelis (December 3, 2025), Retro is raising a $1 billion Series A at a <strong>$5 billion valuation</strong> despite having no clinical data in hand. Retro also published joint work with OpenAI showing that <strong>GPT-4b micro</strong> — a protein-design model — engineered modified Sox2 and Klf4 variants (“RetroSOX” and “RetroKLF”) with &gt;50× higher pluripotency-marker induction than wild-type Yamanaka factors.</p><p><strong>NewLimit</strong>, co-founded by Coinbase CEO Brian Armstrong, closed a $130M Series B in May 2025 (Kleiner Perkins lead) then added another $45M in October 2025 with Eli Lilly participating directly — the first time a top-three pharma has put real money into a longevity startup. Valuation reached $1.6B. Lead program: epigenetic reprogramming of liver hepatocytes for alcohol-related liver disease, with mRNA-delivered transcription factor cocktails.</p><p><strong>Senolytics setback.</strong> <strong>Unity Biotechnology</strong>’s Phase 2b ASPIRE trial of UBX1325 (foselutoclax, a BCL-xL inhibitor) for diabetic macular edema <strong>missed its primary endpoint in March 2025</strong>, despite a positive Phase 2 BEHOLD readout published in <em>NEJM Evidence</em> in April 2025 (“Safety and Efficacy of Senolytic UBX1325 in Diabetic Macular Edema,” Klier et al.). The stock fell ~30% and <strong>shareholders voted to liquidate the company in September 2025</strong>. CEO Anirvan Ghosh told <em>Longevity.Technology</em>: “At this point, we have not hit the mark, and hence we see the broad reactions in the market. But my hope for the field would be that we learn from this — to better pick the right indications, design the right studies, and continue the path forward.”</p><p><strong>Rapamycin’s real data.</strong> The <strong>PEARL trial</strong> (Participatory Evaluation of Aging with Rapamycin for Longevity; NCT04488601) — crowdfunded via Lifespan.io and run by AgelessRx — published in <em>Aging</em> in April 2025 (Moel et al., DOI 10.18632/aging.206235). 114 completers, 48 weeks, 5 or 10 mg compounded rapamycin weekly. Topline: safe, no serious adverse events vs. placebo, modest improvements in lean mass and self-reported quality of life — but the authors acknowledge compounded rapamycin was ~3.5× less bioavailable than commercial formulations, meaning subjects effectively received lower doses than labeled. PEARL is the largest rigorous trial of rapamycin in healthy aging humans to date and confirms safety; it does not yet demonstrate lifespan benefit.</p><p><strong>Metformin.</strong> The <strong>TAME trial</strong> (Targeting Aging with Metformin), Nir Barzilai’s long-stalled flagship, has been partly absorbed into ARPA-H. Barzilai told Lifespan.io in 2025 that “two major trials will come out of this,” including an Eli Lilly TAME-style study using a GLP-1 agonist instead of metformin. As of May 2026, no published TAME results.</p><p><strong>Dog longevity goes regulatory.</strong> <strong>Loyal</strong>’s LOY-002 (a daily oral pill targeting metabolic dysfunction in dogs ≥10 years and ≥14 lbs) hit two of three required FDA technical sections — RXE (Reasonable Expectation of Effectiveness) in February 2025 and <strong>Target Animal Safety in December 2025</strong>. The STAY pivotal trial, completed enrollment July 2025 with 1,300 dogs at 70 veterinary clinics, is the largest clinical trial in veterinary history. CEO Celine Halioua said in a January 2026 release the company is “on track to apply for XCA [Expanded Conditional Approval] next year,” meaning <strong>the first FDA-cleared lifespan-extension drug for any species could arrive in late 2026 or 2027</strong>. This matters beyond dogs: it establishes a regulatory pathway and an FDA-validated endpoint for “healthspan extension” that could be ported to human geroscience trials.</p><p><strong>Bryan Johnson and the citizen-scientist movement.</strong> Bryan Johnson’s Blueprint protocol — 111 supplements/day, 33,000+ biomarkers tracked, ~$2M/year spend — went mainstream in 2025 via the Netflix documentary <em>Don’t Die: The Man Who Wants to Live Forever</em>. Johnson reports a DunedinPACE score of ~0.69 (aging at 69% of normal rate) and claims epigenetic age reversal of 5.1 years across multiple TruDiagnostic clocks. He has publicly <strong>abandoned plasma exchange</strong> (“his team’s data analysis concluded that the procedure did not yield a measurable benefit”) and <strong>discontinued growth hormone</strong> after side effects (intracranial pressure, elevated glucose). In 2025 he launched Blueprint Biomarkers — a $365/year test running ~140 biomarkers on the same TruDiagnostic TruAge platform — explicitly commodifying Blueprint at “less than 1% of the cost.” The American Council on Science and Health and others correctly note this is a case study, not a clinical trial.</p><p><strong>Biomarkers.</strong> <strong>DunedinPACE</strong> (Belsky et al., <em>eLife</em> 2022) has become the field’s de facto pace-of-aging clock, validated in &gt;65 cohorts across &gt;17 countries and &gt;300 publications. A 2025 <em>Nature Aging</em> paper (Sui et al., DOI 10.1038/s43587–025–00897-z) introduced <strong>DunedinPACNI</strong>, a derivative that estimates DunedinPACE from a single structural brain MRI — opening up retrospective analysis of every neuroimaging cohort ever collected. A <em>Nature Communications</em> paper in 2025 (Article 11164) compared 14 epigenetic clocks across 174 disease outcomes; <strong>GrimAge and DunedinPACE remain the most predictive</strong> for incident morbidity and mortality, while the original Horvath clock has fallen behind for clinical applications.</p><p><strong>Framework.</strong> López-Otín, Blasco, Partridge, Serrano, and Kroemer’s “Hallmarks of Aging: An Expanding Universe” (<em>Cell</em>, January 19, 2023; PMID 36599349) updated the 2013 nine hallmarks to <strong>12</strong>, adding disabled macroautophagy, chronic inflammation (“inflammaging”), and dysbiosis. This is now the de facto target taxonomy for every longevity biotech.</p><h3>2. Organ Replacement &amp; Regeneration</h3><p>This is where 2024–2025 produced more clinical drama than the entire prior decade.</p><p><strong>Pig kidneys.</strong> The eGenesis 69-edit pig (Yucatan minipig; 59 PERV-inactivation edits + 10 immunological edits) has now been transplanted into three living US patients at Massachusetts General Hospital:</p><ul><li><strong>Rick Slayman</strong> (March 16, 2024) — died <strong>52 days post-transplant</strong> on May 11, 2024 of an unrelated cardiac event; the kidney was functioning and showed no rejection at autopsy.</li><li><strong>Tim Andrews</strong> (January 25, 2025) — survived <strong>271 days</strong>, a world record, until the kidney was explanted October 23, 2025; he threw out the first pitch at Fenway Park in June 2025 and is back on dialysis. Surgeon Tatsuo Kawai: “This second xenotransplant provides us with another excellent opportunity to learn how we can make genetically-edited pig organs a viable, long-term solution for patients.” Andrews himself: “I liken the trial to going to the Moon. I am just one of the people on this journey that suffered pain, and health issues, along with emotional and personal grief to move the program forward.”</li><li><strong>Bill Stewart</strong> (June 14, 2025) — alive and doing well as of late 2025.</li></ul><p>The Revivicor/United Therapeutics 10-edit pig (the “UKidney”) was used in:</p><ul><li><strong>Lisa Pisano</strong> (NYU, April 12, 2024) — first patient with a mechanical heart pump (LVAD) and a pig thymus co-transplant; kidney removed at 47 days due to inadequate LVAD flow; died July 7, 2024.</li><li><strong>Towana Looney</strong> (NYU, November 25, 2024) — <strong>130-day survival</strong> (then a world record), explanted April 4, 2025 after rejection following a reduction in immunosuppression to treat an unrelated infection. NYU Langone’s Robert Montgomery: “Towana Looney’s genetically engineered pig kidney functioned well for over four months, and she was able to enjoy life without dialysis for the first time in nine years… What triggered the rejection episode after a long period of stability is being actively investigated, but it followed a lowering of her immunosuppression regimen to treat an infection unrelated to the pig kidney.”</li></ul><p><strong>The first regulated xenotransplant clinical trial is now underway.</strong> United Therapeutics’ EXPAND study (NCT06878560) — a Phase 1/2/3 multicenter open-label trial of UKidney in ESRD patients aged 55–70 — performed its <strong>first transplant on November 3, 2025 at NYU Langone</strong>. Initial cohort: 6 recipients; primary endpoint at 24 weeks; eventual BLA target. eGenesis received its own FDA IND clearance on September 8, 2025 for a 30-patient ESKD trial with EGEN-2784.</p><p><strong>Pig hearts</strong> (UMD, Revivicor): David Bennett (Jan 2022, 60 days) and Lawrence Faucette (Sept 2023, 40 days). No new cardiac xenotransplants have been attempted in humans in 2024 or 2025 — the field is regrouping after both early cases ended in antibody-mediated rejection.</p><p><strong>Pig livers.</strong> Three Chinese cases pushed pig-liver xenotransplantation from theory to data in 2024:</p><ul><li><strong>Xijing Hospital (Air Force Medical University), March 10, 2024</strong>: First heterotopic auxiliary pig-liver transplant into a brain-dead 50-year-old; 6-gene-edited Bama miniature pig from Clonorgan; functioned 10 days. <strong>Published in <em>Nature</em>, March 26, 2025</strong> (Tao et al., DOI 10.1038/s41586–025–08799–1).</li><li><strong>Anhui Medical University, May 17, 2024</strong>: First pig-liver xenotransplant into a living patient (71-year-old man with hepatocellular carcinoma); 10-gene-edited Bama pig; auxiliary transplant; <strong>liver removed at day 38</strong>; patient died at 171 days. Published in <em>Journal of Hepatology</em>, 2025.</li><li><strong>Xijing Hospital, January 2025</strong>: First full orthotopic pig-liver replacement (10-hour operation, brain-dead recipient).</li></ul><p><strong>3D bioprinted organs.</strong> United Therapeutics + 3D Systems continue to advance the “Print to Perfusion” platform — 3D-printed human lung scaffolds with 44 trillion voxels, 4,000 km of pulmonary capillaries, and 200 million alveoli, demonstrating gas exchange in animal models. CEO Martine Rothblatt’s stated goal remains “human trials in under five years” (from 2022, so roughly 2027). Kidney and liver scaffolds are in development. A July 2025 industry report cited a $250M UTHR investment in lung-tissue bioprinting; Organovo reported successful preclinical 3D-bioprinted kidney models in August 2025. Honest caveat: no full 3D-printed solid organ has yet been transplanted into a human, and most experts consider the realistic timeline 10+ years.</p><h3>3. Digital Immortality</h3><p><strong>Brain-computer interfaces</strong> had a clearly transformative year. <strong>Neuralink</strong> had 12 implanted patients by September 2025 with 15,000+ cumulative device-hours; the first UK patient (“Paul”) was implanted at UCL on October 28, 2025. First patient Noland Arbaugh, 18 months in, uses the device ~10 hours a day. Neuralink’s Blindsight visual-cortex implant received FDA Breakthrough Device Designation in June 2025. The N1 implant has up to 3,072 electrodes — the highest channel count of any chronic human BCI.</p><p><strong>Synchron</strong> completed its COMMAND early feasibility study (6/6 patients meeting the primary safety endpoint at 12 months), demonstrated Apple Vision Pro and iPad control (the latter announced August 2025), and raised a $200M Series D in November 2025 (bringing total funding to $345M). Synchron is the furthest along the FDA pathway for chronic communication BCIs, with a pivotal trial targeting 2026.</p><p><strong>Precision Neuroscience</strong> (founded by former Neuralink co-founder Ben Rapoport) holds FDA 510(k) clearance for its Layer 7 cortical interface for temporary use (up to 30 days) and is moving toward chronic implants. 2025–2026 disclosed BCI venture investment exceeded $1.6B, including Neuralink’s $650M Series E at a $9B valuation (June 2025), Merge Labs ($252M seed, January 2026), and Science Corporation ($230M Series C, March 2026).</p><p><strong>Connectomics</strong> scaled by an order of magnitude. The <strong>MICrONS Consortium</strong> (Allen Institute, Princeton, Baylor) published “Functional connectomics spanning multiple areas of mouse visual cortex” in <em>Nature</em> on April 9, 2025 (DOI 10.1038/s41586–025–08790-w) — a 1 mm³ cube of mouse visual cortex with <strong>~84,000 neurons and over 500 million synapses</strong> reconstructed, with simultaneous functional recordings during natural-video stimulation. This is the largest mammalian connectome yet, ~20× the complexity of the complete fruit-fly brain (FlyWire, 2024). Princeton’s Sebastian Seung: “The connectome is the beginning of the digital transformation of brain science… With a few keystrokes you can search for information and get the results in seconds. Some of that information would have taken a whole Ph.D. thesis to get before.” Whole-mouse-brain connectomics is the explicit next target; researchers told CNN they expect to know within 3–4 years whether it’s feasible.</p><p><strong>Whole brain emulation</strong> as a route to digital immortality remains speculative. The hard problem of consciousness — whether functional equivalence implies subjective experience — is unresolved; Seung himself, asked about whether a simulated brain would be conscious, deflected: “I don’t have any more authority to make a statement on that than you do. But when people say, ‘I don’t believe a simulation of a brain would be conscious,’ then I say, ‘Well, how do you know you’re not a simulation?’”</p><p><strong>Griefbots and AI replicas</strong> (HereAfter AI, StoryFile, Replika, Sensay) are a fast-growing commercial market with no regulatory framework; “personality capture” via large language models fine-tuned on a person’s writing/audio has become consumer-grade. None of these constitute uploading in any technical sense — they are stylistic simulacra.</p><h3>4. Cryonics &amp; Preservation</h3><p>A small, financially modest, but quietly active field. As of mid-2025, <strong>~500–650 humans globally</strong> have been cryopreserved:</p><ul><li><strong>Alcor</strong> (Scottsdale, AZ): ~250 patients, ~1,500 members, $220K whole-body / $80K neuro.</li><li><strong>Cryonics Institute</strong> (Michigan): 276 patients, ~2,000 members, $28K (cheapest major option). Added a Field Cryoprotection program with Suspended Animation in April 2025 for $18K.</li><li><strong>Tomorrow Bio</strong> (Berlin): The European entrant, raised €5M in May 2025 for US expansion; first European vitrifications in 2024; €160M+ in total contracts; operates dedicated cryo-ambulances.</li><li><strong>Southern Cryonics</strong> (Australia): 4 patients preserved by 2025.</li></ul><p>Total whole-body cryonics market 2025: roughly <strong>$6–11M annually</strong>. Annual preservations across all providers run 30–40.</p><p><strong>Aldehyde-stabilized cryopreservation</strong> (Nectome’s signature approach; sometimes called “fixation”): Oregon Brain Preservation has shifted to a fixation-based protocol; an arXiv preprint from July 2025 surveyed 22 biostasis practitioners on revival biomarkers and obstacles. The Brain Preservation Foundation’s Large Mammal Brain Preservation Prize was awarded in 2018 for ASC; <strong>no human revival from any preservation method has occurred</strong>, and none is technically plausible with current technology.</p><p><strong>Organ cryopreservation</strong> (a separate, much more clinically relevant field) is making real progress at companies like X-Therma and Hypothermosa, working on isochoric and nanowarming approaches to make donor organs storable for days rather than hours.</p><h3>5. Other Frontier Approaches</h3><p><strong>AI-driven drug discovery</strong> for aging is now table stakes. <strong>Insilico Medicine</strong> raised a heavily oversubscribed $110M Series E led by Value Partners Group on March 13, 2025 (closing at ~$123M by June 2025 per CEO Alex Zhavoronkov: “This round… was heavily oversubscribed, drawing exceptional interest from prominent investors”), with 20+ preclinical candidates and a fibrosis drug (Rentosertib) in Phase 2. The OpenAI–Retro Biosciences GPT-4b micro collaboration (early 2025) produced engineered Yamanaka factors with &gt;50× higher reprogramming efficiency. <strong>Gero Pte. Ltd.</strong> signed a deal with <strong>Chugai Pharmaceutical</strong> (Roche) announced July 7, 2025: undisclosed upfront + up to ~$250 million in milestone payments + sales royalties, with total potential value exceeding $1 billion. Chugai President &amp; CEO Dr. Osamu Okuda: “By combining Gero’s target discovery technology with Chugai’s drug discovery technologies, we will accelerate the creation of innovation.”</p><p><strong>Nanotechnology and medical nanobots</strong> — Robert Freitas’s nanomedicine vision, diamondoid mechanosynthesis — remain entirely theoretical. No medical nanobot has entered clinical trials. Kurzweil’s frequent invocation of nanobots as the route to LEV is, at this writing, science fiction.</p><p><strong>Hevolution Foundation</strong> (Saudi Arabia, founded 2022) has deployed roughly <strong>$400 million</strong> across 230+ research grants, ~200 grantees, 25 strategic partnerships, and 4 portfolio biotech companies. CEO Mehmood Khan disclosed at the Global Healthspan Summit 2025 in Riyadh that 2024 global healthspan-science funding nearly doubled to <strong>$7.33 billion</strong>, with average deal sizes up 77% YoY. Khan: “Our ability to extend the number of years people live in good health is one of the defining challenges of our time.” Hevolution co-sponsors the <strong>$101M XPRIZE Healthspan</strong>, which announced 100 semifinalists in May 2025 and will down-select to 10 finalists in 2026, each receiving $1M for full-scale human trials.</p><h3>6. Theoretical &amp; Future Outlook</h3><p><strong>LEV Foundation’s Robust Mouse Rejuvenation Study (RMR-1)</strong>, run by Aubrey de Grey’s team at Ichor Life Sciences in Syracuse, NY, completed in February 2025. 1,000 C57BL/6J mice, treatment started at ~19 months (≈ human age 60), four interventions in combination: rapamycin (42 ppm enteric-coated chow), Galactose-conjugated Navitoclax senolytic, intranasal AAV-mTERT telomerase gene therapy, and young-donor hematopoietic stem cell transplantation. Final readout: a <strong>“qualified win”</strong> — additive (not synergistic) extension of <strong>mean lifespan but not maximum lifespan</strong>, with median extension of ~4 months in the best combination group. The “robust” 12-month target was missed. Clear sex dimorphism: females responded mostly to rapamycin; males showed real synergy when damage-repair interventions were stacked on rapamycin. <strong>RMR-2</strong> is in pilot phase as of late 2025 with 8 interventions (rapamycin + running wheels as universal baseline, cyclic damage-repair dosing, MSCs replacing HSCT, planned addition of OSK partial reprogramming, GDNF, IL-11 inhibition); main phase ~mid-2026, results ~2028. de Grey, as of 2025: “I believe there’s a 50% chance of reaching [LEV] within about 12–15 years from now” — putting his point estimate at ~2036.</p><p><strong>Ray Kurzweil</strong>’s <em>The Singularity Is Nearer</em> (Viking, June 2024) maintained his 2029 AGI prediction and the 2045 singularity, with <strong>LEV “by 2030 for the diligent and well-connected, 2030s for everyone.”</strong> At an October 2025 MIT lecture (“Reinventing Intelligence,” Linde Music Building), Kurzweil reiterated: “These incredible breakthroughs are going to lead to what we’ll call longevity escape velocity.” Critics note Kurzweil’s track record on hardware/compute is excellent, his record on biology timelines less so.</p><p><strong>Aging-as-disease.</strong> WHO’s ICD-11 includes “old age” (MG2A, with the “ageing-related” extension code XT9T) but does not classify aging itself as a treatable disease. No major regulator (FDA, EMA) has accepted aging as a primary indication; this is why every credible aging biotech still has to choose a disease indication (NAION for Life Bio, DME for Unity, MASH for Life Bio’s ER-300, alcohol-related liver disease for NewLimit).</p><p><strong>Public-figure backers.</strong> Sam Altman ($180M into Retro), Jeff Bezos and Yuri Milner (Altos), Larry Ellison (Ellison Medical Foundation, $1B+ historical), Peter Thiel (Methuselah Foundation, multiple bets), Brian Armstrong (NewLimit), and Saudi Arabia (Hevolution). VitaDAO and DeSci movements have funded ~20 small longevity projects with on-chain governance — a small but ideologically distinct stream.</p><h3>7. Critical &amp; Skeptical Perspectives</h3><p>The honest read of 2024–2026 evidence is that we are in the <strong>longevity dividend</strong> era — adding healthy years at the margin — not the <strong>immortality</strong> era. The failed Unity ASPIRE trial, the death of every pig-organ recipient so far, the modest 4-month RMR-1 result, and the lack of any FDA-approved geroprotective drug for humans all argue for calibrated optimism. Andrew Steele’s annual review for The Longevity Initiative (published January 6, 2026, titled “The business of longevity in 2025: big bets amid biotech bust”) put it bluntly: “Longevity fundraising has fared similarly to the rest of biotech overall, which hit both its lowest levels of investment since 2019 in absolute terms, and its lowest levels as a fraction of overall startup investment in a decade.” Outside of Retro’s $1B mega-round, sector capital is unusually thin.</p><p><strong>Inequality and access concerns</strong> are mounting. Bryan Johnson spends $2M/year on Blueprint. Tomorrow Bio costs €220K plus €55/month. Even the cheapest credible epigenetic clock test costs ~$300. Without aggressive price-curve effects (Loyal’s $365/year biomarker platform is a hopeful sign), 21st-century longevity tech risks becoming a status good for the top 1%. The Methuselah Foundation’s stated mission — “make 90 the new 50” — is far more honest than “immortality” framing.</p><p><strong>Ethics of digital immortality.</strong> Replika and Sensay-style “griefbots” raise active questions about consent, manipulation, and grief processing; no jurisdiction has comprehensive regulation. The hard problem of consciousness remains: even a perfect connectome simulation may not entail subjective experience.</p><h3>Recommendations</h3><p>For a tech-savvy reader trying to actually act on this:</p><ol><li><strong>Track DunedinPACE and GrimAge, not chronological age.</strong> TruDiagnostic’s TruAge COMPLETE test (~$300, used by Johnson) gives the best validated battery. Retest every 6 months. These are the only biomarkers with peer-reviewed evidence of responding to intervention.</li><li><strong>Boring beats heroic.</strong> The interventions with the strongest evidence (mostly from animal models with some human data) remain: caloric/protein moderation, Zone 2 cardio + resistance training, sleep regularity, low-dose weekly rapamycin (with a clinician familiar with PEARL-trial dosing), and avoiding metabolic dysfunction. The exotic stack (plasma exchange, GH, follistatin gene therapy) has either no human data or — in Johnson’s own case — negative data.</li><li><strong>If you have access, watch the Life Biosciences ER-100 Phase 1 readout</strong> (likely H2 2026 or 2027) as the canary for whether partial reprogramming works in humans. If ER-100 shows benefit in NAION, expect a cascade of Yamanaka-factor IND filings.</li><li><strong>Watch the Loyal LOY-002 final FDA decision</strong> (likely 2026–2027). The first FDA-approved lifespan-extension drug for any species will reset what is considered scientifically and regulatorily possible.</li><li><strong>Watch the EXPAND trial 24-week endpoint</strong> (~May 2026) for the first regulated read on whether xenotransplant kidneys can match human-donor outcomes. This is the metric to follow if you care about organ replacement.</li><li><strong>For BCIs, the next 18 months bring Synchron’s pivotal trial, Neuralink’s expansion to 20–30 patients, and likely the first BCI commercial PMA submission.</strong> Treat any consumer-BCI marketing before then as marketing.</li></ol><p><strong>Benchmarks that would change these recommendations:</strong></p><ul><li>A Phase 2 partial-reprogramming trial showing &gt;2-line BCVA improvement in NAION → reprogramming becomes a credible whole-body strategy by ~2030.</li><li>RMR-2 showing &gt;12-month lifespan extension in mice → LEV timelines tighten by 5 years; de Grey’s 2036 prediction becomes consensus rather than fringe.</li><li>Any xenotransplant patient surviving &gt;2 years with normal kidney function → organ shortage becomes a solved problem within a decade.</li><li>An FDA approval that uses “aging” or “healthspan” as a primary endpoint → entire field unlocks; expect a Cambrian explosion of geroscience trials.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f2cd50dd331d" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>