<?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 Nate McGuire on Medium]]></title>
        <description><![CDATA[Stories by Nate McGuire on Medium]]></description>
        <link>https://medium.com/@natemcguire?source=rss-9889a2a5666c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*O2dE3dXGKL3v-HcmAVx2bQ.png</url>
            <title>Stories by Nate McGuire on Medium</title>
            <link>https://medium.com/@natemcguire?source=rss-9889a2a5666c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 21:01:26 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@natemcguire/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[Stop building software factories. Give your agents an inbox.]]></title>
            <link>https://medium.com/@natemcguire/stop-building-software-factories-give-your-agents-an-inbox-446c8a0a2754?source=rss-9889a2a5666c------2</link>
            <guid isPermaLink="false">https://medium.com/p/446c8a0a2754</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Nate McGuire]]></dc:creator>
            <pubDate>Sun, 27 Sep 2026 12:07:23 GMT</pubDate>
            <atom:updated>2026-09-27T12:07:23.005Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Tasks, messages, and file reservations are enough to get a team working. Abandon the factory, </em><a href="https://github.com/natemcguire/agent-inboxes/"><em>try agent-inboxes instead.</em></a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*al2dJWvx1PcEqtaDIA8PaQ.png" /><figcaption>Agents like email too.</figcaption></figure><p>I tried building my own software factory, in a few different environments. The agents kept finding work to do on the internal mechanations of the factory instead of doing the work I needed done. Boy it’s fun though.</p><p>They were working on the to-do app instead of the to-do.</p><p>There’s a famous story about Elon Musk at <a href="https://www.tesla.com/">Tesla</a> — they had over-relied on robots and a “crazy, complex network of conveyor belts” that frequently broke down and caused severe bottlenecks. Elon Musk admitted that excessive automation was a mistake, tweeting: <em>“Yes, excessive automation at Tesla was a mistake. To be precise, my mistake. Humans are underrated.”</em> [<a href="https://www.theguardian.com/technology/2018/apr/16/elon-musk-humans-robots-slow-down-tesla-model-3-production">1</a>, <a href="https://www.cbsnews.com/losangeles/news/how-and-why-elon-musk-replaced-robots-with-people/">2</a>, <a href="https://globalnews.ca/news/4149323/elon-musk-tesla-robots-humans-underrated/">3</a>].</p><p>The software factory has the same problem. You’re automating for automation sake. For that dopamine when you hit deploy. But are you actually building anything?</p><p>That is what I care about. Planning, delegation, and process maintenance can all look like progress in a terminal. An agent can produce a considerable amount of activity while the actual task barely moves. Adding an agent to supervise that activity gives you another agent that can get caught in the same loop.</p><p>What helped was giving the agents defined tasks and a way to talk directly to each other. A task tracker, accessible through an API, records the work and who owns it. A mailbox carries questions and handoffs. File reservations let cooperating agents avoid editing the same thing at the same time.</p><p>That combination gets much closer to the working team I wanted. Each part answers a concrete question: what am I doing, who do I need to talk to, and what can I safely work on? I do not need an agent management hierarchy to answer those questions.</p><p>I’m open-sourcing agent-inbox because this is the one part of my many experiments that actually seemed useful.</p><p><a href="https://github.com/natemcguire/agent-inboxes/">View on Github</a></p><p><em>“It helped prevent multiple merge conflicts and release timing was greatly improved” — Claude</em></p><h3>An agent is a strange pipeline stage</h3><p>In our thinking about agents, many of us seem to be treating them like a software build step, or a component of the software itself, rather than what it is — a very good programmer who’s hasn’t slept for a week. I know plenty of those. It can get a lot done. It can also spend an impressive amount of effort pursuing an idea you never asked it to pursue.</p><p>A build step has a contract you can make fairly narrow: run this command against these inputs and produce these artifacts. An agent is choosing commands, interpreting failures, and deciding whether the original plan still makes sense. You can put that inside a pipeline node, but the box hides most of the interesting behavior and utility.</p><p>A planner-to-implementer-to-reviewer pipeline assumes useful boundaries in advance. During real work, the agent implementing a change may discover that the plan is wrong. It needs to resolve that discovery with whoever has the relevant context. Routing the question through a hierarchy adds work to an interaction that could be one message.</p><p>Sounds familar to me, kind of like a team of human engineers.</p><h3>Give them tasks with boundaries</h3><p>“Be the backend engineer” tells an agent very little about what I need finished.</p><p>For example: “Add cursor pagination to this endpoint. Preserve the existing response fields. Show that records aren’t skipped or repeated against an unchanged dataset. Coordinate the cursor format with the agent updating the client.” That gives it a problem, constraints, evidence to produce, and a reason to contact someone else.</p><p>The distinction has consequences for the implementation. I want to know which task an agent owns, what it may change, what it is waiting for, and what would demonstrate completion. A permanent title does not answer those questions.</p><p>The same general agent can investigate a bug, implement the fix, and write a regression test. I can still ask a separate session to review the patch. That review is another bounded task; it does not require maintaining a permanent cast of specialist agents.</p><p>This also means the dependency graph can change while the work runs. Suppose the client agent discovers that the proposed cursor cannot support the required sort order. The useful operation is to send that discovery to the agent changing the endpoint, attach the failing example, and agree on the interface before both go further.</p><p>That is the interaction I wanted agent-inbox to make cheap. I still need ownership, useful tests, and a way to intervene. I do not need to predeclare every conversation as an edge in a workflow graph.</p><h3>The factory becomes the product</h3><p><a href="https://medium.com/u/c1ec701babb7">Steve Yegge</a> supplied a remarkably good argument against reusable agent factories while writing about his own. In his account of Gas Town, he says:</p><blockquote><em>“Gas Town was intended to be reusable, but I only ever wound up using it to build itself.”</em></blockquote><p>He describes it getting stuck in endless tinkering after a model change. He subsequently built Wheelhouse, a bespoke system for another project. He still believes in elaborate agent systems. I read his experience as a reason to make reusable orchestration earn a much higher burden of proof. <a href="https://yegge.ai/essays/the-shape-of-things-to-come/">Yegge’s account</a> of building his factory.</p><p>Dan Luu’s <a href="https://danluu.com/ai-coding/">notes on agentic coding</a> are useful here too. He describes orchestrators that did not fit his work or reliably complete it, and using simpler scripts tailored to the problem. He also points out how quickly useful workflow tricks can get absorbed into the general tools. He is not arguing that all orchestration is useless; his examples put considerable weight on testing and feedback.</p><p>Robert “Uncle Bob” Martin describes an even more direct comparison in <a href="https://x.com/unclebobmartin/status/2098744156709441896">Rethinking Harnesses</a>. After months building orchestration and handoff machinery, he tried giving one agent the same work. He reports that work taking his harness three or four hours could be done better by a single Grok instance in about 40 minutes, even though the harness used the same model.</p><blockquote><em>“Apparently the harness is doing the damage.”</em></blockquote><p>He also describes relaxing his constraints while retaining unit tests, mutation testing, and a code-quality target. His conclusion is to reconsider treating agents as components in a software design. This is his reported experiment, not a controlled comparison of every agent architecture. But it asks exactly the question a factory builder should answer: does the machinery improve on giving a capable agent the task directly?</p><p>My bet is that the general coding agent keeps eating the factory. A role or workflow you painstakingly encode today may become something the next agent handles competently from an ordinary instruction. You still own the machinery after its reason for existing disappears.</p><p>An address and a message are a much smaller bet. Better agents can make better decisions with the same inbox.</p><h3>The team needs three things</h3><p>I use independent agents in separate terminals. Subagents inside a single workflow have been harder to follow and have worked less well for the way I work. With separate sessions, I can inspect each conversation, commands, and output, and intervene directly.</p><p>The useful structure around those sessions is small.</p><p><strong>A task tracker holds commitments.</strong> A task needs an owner, a scope, a state, and a way to decide whether it is finished. A Kanban board gives me a view of that; its API lets the agents use the same record. The task survives the terminal session. I should not have to reconstruct ownership from a conversation transcript.</p><p><strong>The mailbox holds conversations.</strong> An agent can ask about an interface, report a blocker, or send a patch for review. Replies stay attached to the question. A task can link to the thread containing a decision without becoming a running transcript itself. The agent doing the work can contact the agent with the answer directly.</p><p><strong>File reservations cover shared edits.</strong> Owning a task does not imply exclusive ownership of every file it touches. Two unrelated tasks can both require the same module. A reservation makes that overlap visible before both agents invest in incompatible edits. A conflict gives them a reason to coordinate or work elsewhere.</p><p>These records describe different things. A task marked in progress does not reserve a file. A delivered message does not complete a task. A file lease expiring does not prove its former holder stopped writing. The current reservations are advisory: cooperating agents respect them; they cannot fence off arbitrary filesystem writes.</p><p>Consider two agents changing an API and its client. Each owns a task. The client agent finds an ambiguity in pagination and sends a question. They agree on the cursor format in the thread. If both need to edit the shared schema, the reservation exposes the collision. One finishes that edit and hands it over. The final task updates point to the commits and integration checks.</p><p>That is an example of the interaction the system should support. There is no need to route every step through a planner or make every question a new subagent assignment.</p><p>It is also cool to watch. A question gets answered, a blocked terminal gets moving, and I did not have to carry the message. The visible activity has a connection to work getting finished.</p><h3>Why reservations when we already have Git?</h3><p>Git combines changes that have already been made. When it cannot reconcile them automatically, it stops for conflict resolution. By then, both agents have spent the effort producing those changes. <a href="https://git-scm.com/docs/git-merge"><em>Git merge docs</em></a><em>.</em></p><p>A file reservation moves the coordination earlier. Before editing, an agent can discover that another session is already working on that file. It can ask what is changing, agree on a handoff, or pick up another task. The expensive disagreement does not have to be implemented twice before anyone notices it.</p><p>At agent coding speed, that timing matters. Two sessions can generate a substantial amount of mutually incompatible work while each believes it is making progress. Asking them to sort it out afterward is more agent work, precisely what the tooling should be removing.</p><p>A clean merge is not enough, either. One agent can change a function’s behavior while another adds a caller that relies on the old behavior. The patches may touch different lines or different files. Git can combine the text successfully while the program is wrong.</p><p>Reservations reduce overlapping edits; messages coordinate the assumptions that cross file boundaries. Git still provides history, integration, and recovery. I want all three, with preventable collisions caught before the code is written.</p><h3>Keep the bookkeeping out of the model</h3><p>This only works if the tools remove more work than they create. A board that demands constant grooming, an inbox full of status chatter, and a reservation ceremony before every trivial operation can reproduce the factory problem with fewer components.</p><p>Code should handle waiting, delivery, duplicate suppression, and lease expiry. Agents should make the decisions that require context: whether a question matters, whether a change is compatible, and whether the evidence is good enough to call a task done.</p><p><strong>I do not need a language model to supervise a timer. I do not need another agent to summarize a task state that an API can return.</strong></p><h3>What does “an inbox” actually require?</h3><p>The useful test is a failure case. Agent A sends a question. The service stores it, but A loses the connection before receiving confirmation. A retries. Meanwhile, B’s terminal exits.</p><p>I want one stored question, available when B returns, with enough identity and thread information to answer it. That requires persistence, retry deduplication, and an address that survives a particular process. A socket between two terminals does not give me those properties by itself.</p><p>There are also several different acknowledgments hiding inside the word “received.” The service stored the message. The client fetched it. The agent read it. The agent agreed to do something. The work passed its checks. Collapsing those into one boolean is how a reliable transport produces unreliable collaboration.</p><p>For a coding handoff, “done” should point to something the recipient can inspect: a commit, an artifact, the checks that ran, and any unresolved issue. A read receipt cannot substitute for that.</p><p>This is where choosing the transport gets interesting. MQTT, AMQP, NATS, and HTTP are real options, but none automatically supplies the whole collaboration model. An inbox is an application you can build over them.</p><p><em>Different messaging protocols have different characteristics — so of course I had to analyze and test them all.</em></p><h3>MQTT 5: a good event transport</h3><p>MQTT gives you brokered publish/subscribe over topics. It is attractive when tools already emit events and several agents need to subscribe to them.</p><p>Its delivery levels matter. QoS 0 permits loss; QoS 1 permits duplicates; QoS 2 provides exactly-once delivery within the protocol exchange. That last guarantee does not make a file edit or deployment happen exactly once. MQTT 5 also has response topics and correlation data for request/response. Offline delivery depends on session and expiry configuration. A retained message is the latest retained value on a topic, not a conversation archive. <a href="https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html">MQTT 5 specification</a></p><p>For an inbox, I would still define recipient identity, thread storage, read state, and how a new session resumes an old agent’s work. Those are manageable additions. If I already operated MQTT, reusing it could be the simplest choice.</p><p>For a few local terminals, introducing a broker and then building the mailbox above it needs to save enough work to justify both.</p><h3>AMQP 1.0: explicit delivery state and flow control</h3><p>AMQP 1.0 gives you links between messaging endpoints, receiver-controlled link credit, and delivery settlement. A receiver can limit how much the sender sends; settlement lets peers resolve the status of a delivery. These are useful primitives when workers have different capacities or disappear mid-exchange. <a href="https://docs.oasis-open.org/amqp/core/v1.0/os/amqp-core-transport-v1.0-os.html">AMQP transport specification</a></p><p>Its message format includes message IDs, reply addresses, and correlation IDs. Outcomes such as accepted, rejected, released, and modified let clients express more than success or a broken connection. <a href="https://docs.oasis-open.org/amqp/core/v1.0/os/amqp-core-messaging-v1.0-os.html">AMQP messaging specification</a></p><p>I mean AMQP 1.0 here, not the different AMQP 0–9–1 protocol. The version matters when choosing clients and reasoning about broker behavior.</p><p>I would consider it for coordination across services where those delivery controls and an existing messaging installation remove work. I would still need to define persistence and retention in that deployment, and what an accepted message means to the application. Settling a delivery cannot certify that an agent’s patch works.</p><h3>NATS: live coordination, with persistence when needed</h3><p>Core NATS offers subject-based publish/subscribe, request/reply, and queue groups that distribute messages among subscribers. Its delivery is at most once. It is useful for live interactions, but an offline agent cannot treat Core NATS alone as a durable inbox. <a href="https://docs.nats.io/learn/core-nats/">Core NATS documentation</a></p><p>JetStream adds persistence and consumers with delivery state. Pull consumers let workers request messages as they have capacity; acknowledgments and redelivery support recovery. <a href="https://docs.nats.io/learn/jetstream/pull-consumers">JetStream documentation</a></p><p>The design question is who should receive the message. “Someone run these tests” fits a worker pool. “Tell the agent changing the API that this field is required” needs a particular recipient. Putting all agents in one queue group would confuse those two interactions.</p><p>NATS would be a serious candidate if I were connecting agents and tooling across machines. I would put a small inbox API above it so the agent deals with senders and threads, while ordinary client code handles subscriptions and recovery.</p><h3>SMTP and IMAP: The OG</h3><p>Actual email deserves a place in this comparison. SMTP provides store-and-forward delivery. IMAP provides mailbox access, message identifiers, and flags such as seen. Offline recipients are an ordinary part of the model. <a href="https://www.rfc-editor.org/rfc/rfc5321.html">SMTP</a>, <a href="https://www.rfc-editor.org/rfc/rfc9051.html">IMAP4rev2</a></p><p>That is a substantial amount of existing machinery to reuse. An internal mail service with a narrow agent client could be a perfectly sensible implementation.</p><p>The cost is operating or integrating with that mail system and translating its behavior into the interaction the agents need. A seen flag still does not tell another worker whether a task was accepted. For my local setup, that integration was more than I wanted to take on. If the infrastructure already existed, the calculation could go the other way.</p><h3>HTTP and SQLite: what I used</h3><p>Agent-inbox’s local implementation is a Python service, an HTTP API, a command-line client, and SQLite. The database stores messages, threads, recipients, and per-recipient read state. Send retries use an idempotency key so a repeated request returns the existing message instead of inserting another one.</p><p>That behavior belongs to the application. HTTP does not make an ordinary POST idempotent or supply durable storage. <a href="https://www.rfc-editor.org/rfc/rfc9110.html">HTTP semantics</a></p><p>A blocking watch command waits for new mail. Hooks can notify active sessions. The waiting happens in code, without requiring the model to burn turns asking whether anything arrived.</p><p>File and resource reservations handle another concrete problem: two cooperating agents trying to change the same thing. These are advisory leases with expiry, not filesystem locks. They remain local even when messages are synced across machines.</p><p>There is code here to maintain. The simplicity comes from its limited job and from fitting the environment I actually use. If I later need distributed durability or broker-level flow control, I should revisit the transport rather than slowly reconstruct a broker inside this service.</p><h3>Agent Inboxes Now Open Source</h3><p>The useful result of my software factory experiment was a smaller coordination layer: defined tasks, direct messages, and reservations around shared work. Agent-inbox supplies the messaging and reservation pieces; it can be paired with a task tracker rather than requiring its own system for managing the whole project.</p><p>My bet is that improving general coding agents will make this approach more useful. A stronger agent can take on a broader task and ask better questions through the same interface. I would rather benefit from that improvement than maintain a growing set of roles and handoff rules around yesterday’s limitations.</p><p>The measure is whether the actual tasks get finished. If an agent spends the afternoon improving the system that tracks the work, I want to notice that and send it back to the work.</p><p>My software factory didn’t work so I stripped it for parts — this is the mailbox.</p><p>Thoughts? Comments?</p><p><a href="https://github.com/natemcguire/agent-inboxes/">https://github.com/natemcguire/agent-inboxes/</a></p><p>Drop into my <a href="/service/mailto:nate.mcguire@gmail.com">inbox</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=446c8a0a2754" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Work Nobody Wanted to Do]]></title>
            <link>https://medium.com/@natemcguire/the-work-nobody-wanted-to-do-06ef9a181d15?source=rss-9889a2a5666c------2</link>
            <guid isPermaLink="false">https://medium.com/p/06ef9a181d15</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Nate McGuire]]></dc:creator>
            <pubDate>Mon, 14 Sep 2026 23:55:31 GMT</pubDate>
            <atom:updated>2026-09-14T23:55:31.870Z</atom:updated>
            <content:encoded><![CDATA[<p><em>What a small website taught me about slow builds, scarce attention, and where AI changes the (economics) game</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hkFAex6EOuuClq5nRD_KGw.png" /></figure><p>I was pulling up one of my sites, <a href="https://purecalculators.com/">purecalculators.com</a>, a small ads/content website I bought a while back, and remembering how much work it took to make the thing easier to work on after I bought it.</p><p>The old system used Notion as its content backend. This blew my mind that anyone would do this but, alas, now I own the site so it was my problem.</p><p>I can see the appeal of starting with a tool people already know. But inheriting a business whose publishing process was built around Notion meant inheriting an unusual way of getting things done. Changing the site took more work than the simple pages suggested.</p><p>We spent a lot of effort trying to work on the site and make updates. The changes we actually managed to get through were cosmetic. The feedback loop was so slow that doing deeper work would have been cost-prohibitive, not to mention just painful — and not fun.</p><p>This is an under discussed part of our current moment. Yes, there is a ton of slop, but there’s also work getting done that wouldn’t have otherwise. That is the part I keep coming back to when I think about AI and software.</p><h3>A slow build can eat the whole project</h3><p>A project you work on around other responsibilities has a different relationship with time.</p><p>You get an opening in the day. You remember something you wanted to fix. You open the repository, work out where you left off, make a change, and start the build.</p><p>Then you wait.</p><p>While you wait, something else needs your attention. By the time you come back, you have to remember what you were checking. If the build failed, you get to make another change and go around again. If the build process takes many minutes, or an hour, you lose track and move on to tighter loops.</p><p>The cost includes all the times you have to reload the project into your own head.</p><p>A slow feedback loop makes a small task difficult to fit into a small window. Eventually, you start reserving the work for a longer stretch of uninterrupted time. Those stretches are scarce. The backlog stays where it is.</p><p>That was the situation with Pure Calculators during those early attempts. We were willing to work on it, and we did try. But every round of making a change and checking the result added to the cost. We could get cosmetic updates through; the more substantial work was hard to justify economically.</p><p>You have to understand where the time goes before you can improve it. Even figuring that out is a task someone has to do.</p><h3>Before the work, there is more work</h3><p>I also remember how hard it was to find good engineers who wanted to take on the backend work.</p><p>I understand why. An existing piece of software comes with someone else’s decisions, unfamiliar scripts, and behavior you need to preserve. There is a learning curve before you can be useful. The project has to make sense for the person taking it on: the money, the commitment, the opportunity cost.</p><p>From my side, there was work involved in finding that person, explaining the system, establishing trust, and keeping things moving. For a small business, that overhead can be substantial relative to the actual change you need.</p><p>The website mattered to me. That did not automatically make its maintenance an attractive assignment for somebody else. I’m sure many other founders and engineering leaders know the feeling.</p><p>Finding someone capable was only part of the problem. Even with help, the slow feedback loop made useful progress expensive. The work had to make economic sense for a small advertising-supported website, and the deeper changes did not.</p><h3>AI does not need an interesting assignment</h3><p>Once AI started becoming available, I ended up doing the conversion myself, by hand. I moved the site to a more sensible backend.</p><p>That is the sequence that matters: we had struggled to get beyond cosmetic updates, and I eventually took on the deeper work myself. Looking back at that experience is what makes me interested in how AI changes the economics of maintaining a small website.</p><blockquote>I can ask a coding agent to investigate why a small change requires so much waiting. I do not have to make the assignment sound like an exciting career opportunity first.</blockquote><p>I used to need someone who was both good enough to do the work and interested enough to take it on. For some of that work, the second requirement is starting to matter less.</p><p>The first requirement still matters enormously. An agent can enthusiastically misunderstand a system. It can propose a cleanup that removes something important. Its willingness to attempt a task tells me very little about whether the result is correct.</p><p>But I can approach the backlog differently. A task that previously required finding someone and arranging a handoff can begin with a focused investigation in the repository I already have.</p><p>An agent does not automatically make a slow build faster. But it can help carry an investigation through the repeated changes, builds, and checks that made the old process so expensive to work through. I can ask for one bounded improvement and evaluate the result.</p><p>It is also exactly the kind of work that is easy to put off when the project lives at the edge of your day.</p><h3>Make the next change easier</h3><p>The appeal extends to the rest of the maintenance backlog: documenting a confusing script, checking an outdated dependency, or adding tests around calculator behavior that must survive a refactor.</p><p>I would want those tests to check answers against independently established examples. A test that repeats the same mistake as the implementation offers very little reassurance.</p><p>The same standard applies to build improvements. A faster build is useful only if the site still works properly afterward.</p><p>Choosing the task, setting those constraints, and reviewing the evidence remain my responsibility. So does deciding whether a change is worth making at all. A small website can acquire plenty of new complexity in the name of cleaning up the old complexity.</p><p>The question I care about is whether a useful change can be completed and checked at a cost the business can justify. Generating more code is only valuable if it helps us get there.</p><p>And I would put work that makes future changes easier near the top of the list. If the development process fits into the time I actually have, more of the rest of the backlog becomes possible.</p><p>That is what makes this interesting to me. There are useful little businesses sitting behind awkward systems. Their owners know things need attention. Sometimes the obstacle is finding help; sometimes it is a process that turns every small improvement into an afternoon.</p><p>With many side projects and small businesses online — the deeper work was too expensive to justify through that slow feedback loop. Once AI started becoming available, I ended up doing the backend conversion myself, by hand. That experience shapes what I want from the tools now: help getting necessary work all the way through to a useful result.</p><p>Fixing the site had always been worth doing. It just hadn’t been worth what it cost to get it done.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=06ef9a181d15" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Flovatar Genesis Pack Drop Date and Details]]></title>
            <link>https://medium.com/@natemcguire/flovatar-genesis-pack-drop-date-and-details-9194ec212558?source=rss-9889a2a5666c------2</link>
            <guid isPermaLink="false">https://medium.com/p/9194ec212558</guid>
            <category><![CDATA[dapper-labs]]></category>
            <category><![CDATA[flow-blockchain]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[bitcoin]]></category>
            <category><![CDATA[nft]]></category>
            <dc:creator><![CDATA[Nate McGuire]]></dc:creator>
            <pubDate>Mon, 29 Nov 2021 22:40:50 GMT</pubDate>
            <atom:updated>2021-11-29T22:50:46.438Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>NFTs on Flow: Flovatar Genesis Pack Drop Date and Details</strong></h3><p><em>This </em><a href="https://medium.com/@luca_flovatar/flovatar-genesis-pack-drop-date-and-details-8931e41908dd"><em>article</em></a><em> was originally written by Luca, the founder of </em><a href="http://flovatar.com/"><em>Flovatar.com</em></a><em>, a new NFT project built on Flow. Genesis version is dropping December 16th. Read on for innovations in NFT development, Flow blockchain development, and more.</em></p><p><a href="https://flovatar.com/">Flovatar</a> is the next generation PFP project built on Flow where collectors take the creative lead to shape their own unique characters. Participants will be able to select their favorite combinations across seven main categories with over 3 billion different combinations. No more trade-off between rarity and your own individuality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*TLreI0plLDBbbi3o.png" /></figure><p><strong>Only 9,999 can be minted</strong> as the original series and each of them come with superpowers:</p><ul><li>Commercial rights license to your own Flovatar (swag, toys, or movie deals anyone?)</li><li>1% lifetime royalty for each secondary trade to the original Flovatar creator (you)</li><li>Early access to Stardust token ($DUST) and DAO participation</li><li>Future airdrops, exclusive swag, and many surprises to come</li></ul><h3>Drop date</h3><p>The Genesis public drop will be on <strong>Dec 16th at 9am PT / 12pm ET / 5pm UTC</strong>. As previously <a href="https://medium.com/@luca_flovatar/flovatar-lite-paper-b2a1824e1a45">mentioned in this article</a>, there will be a total of <strong>8,999 packs</strong> available for purchase between the <strong>pre-sale on Dec 14th at 9am PT lasting 24 hours</strong> (for whitelisted members, <a href="http://discord.gg/flovatar">join Discord</a> to learn more) and public sale on Dec 16th.</p><h3>Pack structure and pricing</h3><p>There will be two types of packs available for purchase: Standard and Deluxe. Inside each pack, you will find a single Spark needed to give life to a Flovatar. Where they differ is the number of rarity Boosters and Flobit components.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*U_igASvdCbztE58V.png" /></figure><p>Standard and Deluxe packs will present different chances of getting epic and legendary Boosters and Flobits.</p><p>For a primer on Spark, Boosters, and Flobits, you can <a href="https://medium.com/@luca_flovatar/flovatar-lite-paper-b2a1824e1a45">take a look here</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*htNEQmRyb5bqaVvc.png" /></figure><h3>Price</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QD6hGQIKCMCQ1fac.png" /></figure><p>We wanted to ensure that Flovatar is as accessible as possible to most members, so that is why we decided to keep the Standard pack below the average price of many previous PFP projects, while the Deluxe packs can be quite attractive for collectors. Each pack will contain a single Spark, ensuring a fair distribution for all participants. Please find below additional details on Boosters distribution and the chances for each tier in a pack.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*IegnGkRniHlwNkEs.png" /><figcaption>An example of of our Legendary body traits that can be unlocked with the Rarity Booster</figcaption></figure><p>An example of of our Legendary body traits that can be unlocked with the Rarity Booster</p><h3>Rarity Boosters Distribution</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*4GR7e_Ogb4INqks1.png" /><figcaption>Amount of Rarity Boosters minted and their Pack distribution</figcaption></figure><p>Amount of Rarity Boosters minted and their Pack distribution</p><h3>Boosters Chances</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*PJKmrqC6G6KyH4RI.png" /><figcaption>Chance to find a corresponding rarity tier for each Rarity Booster in the Pack</figcaption></figure><p>Chance to find a corresponding rarity tier for each Rarity Booster in the Pack</p><p>If you ever wonder how we randomized the distribution for all items, here is the <a href="https://github.com/crash13override/flovatar-randomizer/blob/main/RandomizePacks.php">link to the randomize function</a>.</p><h3>Team Allocation</h3><p>As an additional token of fairness, we are not reserving any packs for the team, but simply allocating 200 Sparks for team members and key contributions/advisors. This means <strong>there won’t be any Boosters or Flobit components allocated to internal team members</strong>, ensuring total alignment to maximize value for the community. Of course, each team member will have the freedom to join the drop and purchase packs like every member if they decide to do so (no preferential queuing will be given) or participate in the marketplace.</p><h3>A bit about Flobits</h3><p>Flobits can be added to your favorite Flovatar like accessories or to enrich the story of a character. Differently from the main body components, each Flobit is a unique NFT that can be applied to a Flovatar but also be traded on the Marketplace. They can be viewed like skins that you can change in a video game, but as NFT, with the option of applying multiple Flobits from different categories on the same Flovatar.</p><p>The main categories are:</p><ul><li>Hat</li><li>Eyeglasses</li><li>Accessories<br>(sport items, tools, gadgets and more)</li><li>4th category? (To be revealed post drop)</li></ul><p>Flobits will come in various rarity tiers, with different edition sizes described below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/869/0*3iZrXnt25aqvF1-e.png" /><figcaption>Flobits Rarity tiers</figcaption></figure><p>Flobits Rarity tiers</p><p>Please note that for the Genesis drop, Flobits up to the Legendary tier will be available. The Ultimate Flobits will be announced at a later date after the drop.</p><p>Based on the current number of Flobits already designed, we expect 25k-35k total Flobit NFTs to be minted for this round following the mint size chart presented above for each component. The good news is that all Legendary and Epic components currently available in the Builder will be included in the Genesis drop. However, given the total number of Flobits NFTs in this drop will be capped at 12,798 based on the pack structure mentioned earlier, Rare and Common components will be randomly selected to fill the rest.</p><p>The remaining NFTs will be part of an upcoming Flobits release, where we will also allow the community to submit new accessory proposals. A portion of Flobits will be also used as rewards for various community contests.</p><h3>Flobits Allocation</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/881/0*ea7C8vLYum3V-t2C.png" /><figcaption>Amount of total Flobits minted and their Pack distribution</figcaption></figure><p>Amount of total Flobits minted and their Pack distribution</p><h3>Flobit Chances</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*6c0FpV5SBjXfi0xq.png" /><figcaption>Chance to find a corresponding rarity tier for each Flobit in the Pack</figcaption></figure><p>Chance to find a corresponding rarity tier for each Flobit in the Pack</p><h3>How to Purchase</h3><p><a href="https://blocto.portto.io/"><strong>Blocto wallet</strong></a> will be the main one supported, so we highly suggest anyone create a wallet in advance and load it with the needed FLOW you are planning to spend during the drop. There is a good chance the drop might sell out quickly. Especially for members who have their FLOW token staked, you will probably want to <strong>unstake right away before Tuesday, Nov 30th</strong>, because it takes a two weeks period to be able to withdraw your staked FLOW (just in time for the drop ;))</p><p>We have also integrated with <a href="https://www.moonpay.com/"><strong>Moonpay</strong></a>, which allows you to purchase FLOW with Credit Card or other payment systems (ApplePay, GooglePay) and send the tokens directly to your Blocto wallet. If you wonder which exchanges can be used to get FLOW, feel free to take a <a href="https://coinmarketcap.com/currencies/flow/markets/">look at this link</a> for the one that fits you best.</p><p>Pro tip: To best prepare in advance of the drop, we also suggest visiting our website Flovatar.com a week before the drop in order to connect your wallet and initialize your account to create the Flovatar Collection in your wallet. This will possibly save you some precious time during the drop. When the date comes, you will notice a “Connect” button on the portal that will trigger a simple popup for you to connect. Once initialized your wallet will be able to interact with Moonpay too.</p><h3>Queueing system</h3><p>In order to ensure a good user experience, we implemented a queuing system integration that will create a waiting room approximately 15 minutes before the drop.</p><p>At the time of the drop each person will receive a randomized seat order and we will allow a controlled amount of users every minute to complete the purchase, so that there won’t be any overloads on the system.</p><p>Each user will be able to buy <strong>just one pack at the time. </strong>Once a user completes the pack purchase, if they like to buy another pack, they will be sent to the back of the queue. There is no limit to the maximum number of pack each member can buy, they’d simply go through the queue each time till the drop is sold out.</p><h3>Long Term Sustainability</h3><p>Flovatar was created to be community-driven and much more than a simple PFP project. We are aware that the long-term success of the project hinges on us all working together to bootstrap the best community and ecosystem surrounding the Flovatar Universe. That is the reason why the team decided to allocate the majority of the revenue into the project Treasury that will be mainly used to grow the ecosystem, bootstrap needed components, and support individuals, artists, or entire teams building on top of Flovatar.</p><p>For the Genesis drop, upcoming Flobits drops, and Marketplace transaction fees, 60% of all the revenues will go directly into the project Treasury while only the remaining 40% will be distributed among the core team members. We will make the Treasury wallet public before the drop for transparency. Along the way, it is our goal to empower the community of Flovatar owners to be able to assist us in the roadmap and Treasury management via the upcoming DAO.</p><p>We will need more time to structure the DAO appropriately, in a thoughtful way, that allows large participation without hindering the speed needed for an early-stage project. Rest assured that all Flovatar owners and early adopters will have a voice in it and we are in this together for the long term.</p><h3>Commercial Rights</h3><p>Each Flovatar will grant their owners additional benefits something kin to a full commercial license lease in perpetuity. This means a Flovatar owner can independently pursue collaborations, sponsorships, or commercial deals independently. 90% of any potential revenue from these deals will be retained by the Flovatar owners, while only 10% would need to be distributed back to the DAO to continue the support of the Flovatar ecosystem, further increasing the utility and global awareness for all owners.</p><p>The project entity will retain the full copyright of all Flovatars in order to be able to provide legal support and a united front in case of external attacks or challenges, but in turn, it will lease the commercial right to each owner via the DAO in perpetuity. For transparency, we would like to note that there might be extreme circumstances in the future when the DAO might need to revoke the commercial rights lease from a specific Flovatar owner if the usage of Flovatars were to be redeemed unethical or plainly illegal (e.g. crimes recognized by UN and/or including but not limited to using Flovatars as a symbol of hatred, violence or discrimination). We expect this to never happen, but it is very important to ensure there are processes in place to avoid a few bad actors ruining the entire ecosystem. That said, it will require at least 70% of all votes in the DAO to be cast in favor of the action of this magnitude to take effect. $DUST will play a very important role in the DAO voting and global consensus coordination. We will be sharing more info on this front in the coming months as each process needs to be planned very carefully.</p><h3>Conclusion</h3><p>While we have tried to lock in as many details as possible in this blog, there are a lot of variables to consider. We kindly ask all our community members to be understandable in case we need to make some final edits before the <strong>public drop on Dec 16th</strong>. In case of any major changes, we will ensure to communicate as early as possible across all channels (we highly suggest <a href="http://discord.gg/flovatar">joining us on Discord</a> or <a href="https://twitter.com/flovatar">follow us on Twitter</a> to get the latest updates).</p><p>Now go play with our awesome <a href="https://flovatar.com/builder">Builder</a> and go with the Flow!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9194ec212558" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Time is Money and Money is Time. How Do You Manage Both?]]></title>
            <link>https://medium.com/@natemcguire/time-is-money-and-money-is-time-how-do-you-manage-both-505bff438ebe?source=rss-9889a2a5666c------2</link>
            <guid isPermaLink="false">https://medium.com/p/505bff438ebe</guid>
            <category><![CDATA[branding]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[time-management]]></category>
            <dc:creator><![CDATA[Nate McGuire]]></dc:creator>
            <pubDate>Fri, 15 May 2020 15:24:29 GMT</pubDate>
            <atom:updated>2020-05-15T15:24:29.191Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/600/0*kCtdbFOabFcHqxMt" /><figcaption>Yeahhhhh, about that time sheet….</figcaption></figure><p>Listen. You hate time tracking. I hate time tracking. We all hate time tracking. Unfortunately, in the services business, time tracking is vital to success. Lucky for us, many developers have died on the hill of making the new, best, time tracker! Which means there are a lot to choose from, including ours. 👋</p><p>When you build a time tracker, you quickly learn you don’t really compete with those other time trackers anyway. We’re all just stealing marketshare from the good ole’ spreadsheet or the Award Winning write-on-a-napkin-when-i-remember methodology.</p><p>Teams are really bad at tracking time. A better time tracker isn’t the answer.</p><p>The tricky part isn’t figuring out <em>how</em> to track time. The hard part is compliance from your team and ROI. <em>Why am I tracking time again?</em>. If you can demonstrate the <strong>why</strong>, you can ensure that your team is compliant. I’ve heard so many stories about how of an agency writes their time down on paper, only to submit at the end of the week (or when they remember). 🤯</p><blockquote>Are you profitable so far this year?</blockquote><blockquote>How much did that project cost to deliver?</blockquote><p>You have to be able to answer this for any business, especially</p><p>How you leverage time management can be the difference between profit and loss or your best team member leaving because of burnout.</p><p><strong>Why Track Your Time</strong></p><p>We all track time because time is money. You may bill clients hourly, or flat rate, or some combination — we do both at Mayven. Most people think that tracking time is only about billing time, when really it’s actually about tracking expenses.</p><p>Every person on your team, whether paid hourly or on salary, has an hourly rate that reflects their cost basis (how much you pay them per hour including expenses) as well as a utilization percentage. I observed how this works at my first job, working at the OG Services company, Ernst &amp; Young. Basically you want your utilization rate above 80% — 80% of the hours you or your employees are working ideally are billable to a client or project. If you hit 80% *most* of the time, you will more than cover the cost of that person’s full time work. The other 20% of time is usually spent on internal activities, training, and for us, side projects.</p><p>When the whole team adopts time tracking, everyone has visibility into how much people are working and how long tasks take. Maybe Connor is spinning his wheels on something that may not be high on the project’s priority list. Layer this onto a team with multiple projects and this problem multiplies, distracting you from your goal.</p><p>It’s also a great way to gauge individual workload and prevent burnout. How many times have you looked up from your work and wondered where the hours went? It happens to all of us. Time tracking can help us find a stopping point and disconnect from work. We need this disconnect for our own sanity and for the health of our team.</p><p><strong>Who Tracks Time?</strong></p><p>On our team, everyone tracks time. Primarily we track time to individual projects and tasks so that we can understand where our time is spent. With Buildstack, we’ve automated the process of tying time tracking to your hourly rate, so you can see how much a project costs you to deliver without having to calculate it manually. Automatically showing if you’re profitable, or not.</p><p>There are a couple of things to remember so time tracking doesn’t impede your productivity. If you manage a team, don’t use it to micromanage your team. <a href="https://www.protocol.com/remote-work-boss-tracking-tools">Chris Rothe</a>, co-founder of the computer security firm Red Canary said, “There’s the potential to leverage [time tracking] tools to enforce management when you can’t see over your employee’s shoulders.” I really recommend you not do that, and instead trust your team members to keep track of time, and make sure they enter it on a regular basis.</p><blockquote>Yeahhhhh, about that time sheet….</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/0*kCtdbFOabFcHqxMt" /></figure><p>This brings me to my second point: With any type of work, there’s always a tradeoff between quality and efficiency. If you are constantly thinking about time, you might be compromising your quality — and best work — for speed. I would rather deliver the best product possible than finish it in record time. You owe it to yourself and to your client to do the best possible job, not the fastest.</p><blockquote><strong><em>“Time is money, but also money is money.” — William Gibson</em></strong></blockquote><p>Your team might be the most talented, most passionate lot you’ve ever met, but don’t fool yourself…everyone works to get paid. Whether you’re freelancers, full timers, or contractors, your time is valuable and clients know this. You need to be able to show them the numbers to back up your bids and proposals. Enter billable hours…</p><p>This is time that can be charged to the client for a specific project or task. You can also use this information to see if your fixed fee services are actually profitable. It is important to be as specific as possible with your time tracking, so you can manage your workflow during a project and assess profitability for future projects.</p><p>What is considered <a href="https://www.freshbooks.com/hub/accounting/billable-hours">billable time</a>? Here are a few activities:</p><ul><li>Project planning, research, and timeline development</li><li>Communications, including meetings, phone calls, and emails</li><li>Performing actual the work and any revisions</li><li>Yes, Karen, we have to bill you for the extra 30 page designs</li></ul><p>The Mayven team bills time to specific tasks, within a specific project. There are a few default settings, client calls, for example, but otherwise we attribute time directly to the task we are working on.</p><p><strong>Time Tracking and Communications</strong></p><p>Our distributed team relies on asynchronous communication simply because we work across different time zones. We’ve also found it enables us to have more thoughtful, productive conversations, because we’re not in a rush to reply right away. And sometimes, we just want to share a funny meme or talk about video games — definitely important, but not urgent.</p><p>So what counts when you’re tracking time on projects? How do you keep track when your team is distributed and not on the clock at the same time?</p><p><strong>Try Buildstack’s integrated time tracker for free at </strong><a href="https://buildstack.com/"><strong>buildstack.com</strong></a><strong>.</strong></p><h3>⏱️ to ✌️</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=505bff438ebe" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Remote Work Is Here to Stay — So, Now What?]]></title>
            <link>https://medium.com/@natemcguire/remote-work-is-here-to-stay-so-now-what-1680e8a4398?source=rss-9889a2a5666c------2</link>
            <guid isPermaLink="false">https://medium.com/p/1680e8a4398</guid>
            <category><![CDATA[internet]]></category>
            <category><![CDATA[remote-working]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[coronavirus]]></category>
            <category><![CDATA[covid19]]></category>
            <dc:creator><![CDATA[Nate McGuire]]></dc:creator>
            <pubDate>Thu, 16 Apr 2020 22:54:05 GMT</pubDate>
            <atom:updated>2020-04-16T22:57:14.322Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Remote Work Is Here to Stay — So, Now What?</strong></h3><p>In the early days of the pandemic, non-remote businesses stumbled as they tried to get remote teams up and running. There were technical issues, of course, and they worried about lost productivity. Now we’re a few months in, and a new problem has surfaced: <em>the people</em>. Let me explain…</p><p>For many, the work day BC (Before COVID-19) probably looked like this:</p><ul><li>Get ready, commute to work, listen to a podcast</li><li>Small talk with people walking from the parking garage</li><li>Standing around the coffee maker while you contemplated starting to work</li><li>“Hey you want to grab something from the cafeteria for lunch? I heard the sushi food truck is back”</li><li>OMG SKITTLES IN THE MICRO KITCHEN</li><li>Impromptu cube-side or hallway huddles</li><li>One on one meetings with managers</li><li>After-work happy hours with co-workers</li></ul><p>The common denominator here is <em>people</em> physically together, working and interacting IRL. These activities create the personal connections that help build a company’s culture. In an After Covid world, we have to be intentional about the digital spaces and interactions that we create to bring together our teams.</p><p>Sure, apps like Slack are great for communication. But are they great for the way your team communicates? They may encourage a flurry of DMs, but at the same time become a huge distraction from the actual work.</p><h3>The answers lie in <strong>intentionality and expectations</strong>.</h3><p>Do you really need to jump to reply to every email or message immediately? Or can some communications wait until later, or even a different day?</p><p>At Mayven and Buildstack, we thrive on a mix of synchronous and asynchronous communication. In order for this to work, though, we have to set expectations about what is and isn’t acceptable communication, how fast you should respond, etc, and actually communicate this to the team (most teams forget the last part). By setting expectations, we can then be intentional about our communication and workflows, creating specific spaces for work and for play.</p><p>Aside from important Zoom meetings, team members have total control over how they structure their days. You’re measured on results, not on who you had lunch with. Purposeful communication is more valuable than real-time/all-the-time communication.</p><p>We’re definitely not alone in this thinking. <a href="https://www.owllabs.com/blog/remote-work-enablement">Patrick Cournoyer, COO at Peakon</a>, said</p><blockquote>“We have this inherent trust with all of our employees. We don’t manage based on time or hours in the office. We manage to work outputs as opposed to time.”</blockquote><p><a href="https://www.owllabs.com/blog/remote-work-enablement">Patrick Cournoyer, COO at Peakon</a></p><p>So if a great remote company culture is built on trust, and trust is built on personal connections: How do we create those work happy hour moments online? Is this the Holy Grail — something unattainable, mythical? Nah…we think it’s something that — with a little work — can be easily realized.</p><p>So far our team has had…</p><ul><li>Nate’s bar, where we rotated between Tiki Lounge and Star Wars Canteen</li><li>Friday Pizza Night where we shared tasty pics of our pizza for 24 hours as each of us signed off for the weekend</li><li>Informal roundtables with friends both inside and outside the company. Meeting new people is hard to come by now, and valued heavily</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/540/1*6fmYcVQ_brxXOfss0zU1Zw.gif" /><figcaption>Pineapple belongs on Pizza. I will die on this hill.</figcaption></figure><h3>Now, we’re debating which recipe we should make and also thinking about how we pull off a remote painting and BYOB class. 🍷🎨</h3><p>It may seem like these days are going on forever, but we’re just at the beginning of a major change in the way we all live and work. It’s important to keep your team together and motivated as we navigate these new waters.</p><p>Want to learn more and hear how other organizations are coping? Join our upcoming live panel, where I’ll chat with leaders from Twitch, Bridgepointe, and Patagonia about their perspectives on the shift to digital.</p><p>Join us for <strong><em>Stories from IRL: How the COVID-19 crisis is changing the way we do business </em></strong>live on the <a href="https://www.remoteworkoutpost.com/">Remote Work Outpost</a> on April 21, 2020 at 1 PM PST.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1680e8a4398" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[If Software is Eating the World, your office is also on the menu]]></title>
            <link>https://medium.com/@natemcguire/if-software-is-eating-the-world-your-office-is-also-on-the-menu-56122126f3df?source=rss-9889a2a5666c------2</link>
            <guid isPermaLink="false">https://medium.com/p/56122126f3df</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[remote-working]]></category>
            <category><![CDATA[gitlab]]></category>
            <category><![CDATA[buildstack]]></category>
            <dc:creator><![CDATA[Nate McGuire]]></dc:creator>
            <pubDate>Wed, 04 Mar 2020 22:53:03 GMT</pubDate>
            <atom:updated>2020-03-11T20:58:25.643Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>If ‘Software is Eating the World’, your office is also on the menu</strong></h3><p>March 2020: Companies like <a href="https://www.bizjournals.com/sanfrancisco/news/2020/03/02/salesforce-and-twitter-halt-most-corporate-travel.html">Salesforce</a>, <a href="https://twitter.com/balajis/status/1234764753563766784">Stripe</a>, <a href="https://www.latimes.com/business/technology/story/2020-03-02/twitter-tells-employees-to-work-from-home-amid-coronavirus-fears">Twitter</a>, <a href="https://www.washingtonpost.com/business/2020/02/28/coronavirus-cancel-travel/">Facebook</a>, <a href="https://www.washingtonpost.com/business/2020/02/28/coronavirus-cancel-travel/">Amazon</a>, and others are having their employees cut travel and work from home in reaction to a possible pandemic. Nine years since software started “eating the world” (<a href="https://www.wsj.com/articles/SB10001424053111903480904576512250915629460">Marc Andreessen’s oft-quoted essay</a>), your office is now also on the menu. With companies leaning into distributed work as a precautionary measure, will we be able to get the <strong>remote work genie back in the bottle?</strong></p><p><strong>I’m </strong><a href="https://twitter.com/natemcguire"><strong>Nate McGuire</strong></a><strong>, founder of </strong><a href="https://mayvenstudios.com/"><strong>Mayven</strong></a>, a 100% remote design and engineering agency with over 50 people, where we help Silicon Valley brands and startups scale. To support our team, we also built our agency’s operating system, <a href="https://buildstack.com/">Buildstack</a>, which helps remote services teams run their business. I’ve been working remotely for 10 years and see companies of every size struggle in silence because they don’t realize that they are, in fact, a remote company.</p><p><a href="https://twitter.com/darrenmurph"><strong>Darren Murph</strong></a><strong>, Head of Remote at </strong><a href="https://about.gitlab.com/"><strong>GitLab</strong></a>, the world’s largest and fastest growing all-remote company, joins me in this discussion. He’s worked remotely for 14 years in all stages of remote. In his current role, he works to empower more people and companies to go remote, implement better ways of working, and equip his team with the tools and processes they need to thrive in an officeless environment.</p><p>Here, Darren and I explore strategies to create a remote and distributed culture based on our observations spanning across companies sized from 1 to 1000+.</p><p>When we talk with friends who have to go into the the office, what do they complain about? The commute. People want to be working remotely in a digital economy because it eliminates one of the worst parts about work — the commute. Many employees would work remote in a heartbeat if only the company and culture supported the transition to go remote.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ftenor.com%2Fembed%2F3411842&amp;display_name=Tenor&amp;url=https%3A%2F%2Ftenor.com%2Fview%2Fbruce-almighty-comedy-jim-carrey-frustrated-come-on-gif-3411842&amp;image=https%3A%2F%2Fmedia.tenor.com%2Fimages%2F006338b8018b14dbea5c8410e210922e%2Fraw&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=tenor" width="600" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/0b46110dd8dbfd12b5c696a94174890f/href">https://medium.com/media/0b46110dd8dbfd12b5c696a94174890f/href</a></iframe><p><strong>Remote teams:</strong></p><ul><li>Save overhead on offices</li><li>Have higher worker productivity</li><li>Better voluntary retention</li><li>Are inherently more diverse</li></ul><p>So, how do you create a remote first culture?</p><p><strong><em>Build the foundation: Company size, 0–50</em></strong></p><p>Start by building good habits. When the company is small, it’s important that the foundation is strong as it is the base from which your company’s remote culture will grow. So, it’s most important for founders or early team members to:</p><ul><li>Focus on excellent written communication</li><li>Establish company norms on async versus synchronous communication, and when each is preferred</li><li>Make it easy for new team members to be remote by supplying equipment and providing an office stipend if they don’t have a good spot to work from home</li><li>Document as much as possible about processes and customers in order to onboard new team members quickly and easily</li><li>Find a way to meet in real life! We’ve had big meetups and we’ve also flown around the world to see our team. The point here is to meet, share a meal, and finally see each other in person.</li></ul><p><strong><em>Starting to scale: Company size, 50–1000</em></strong></p><p>As companies grow, especially those that didn’t start remote, there will be opportunities to foster a remote first culture that meaningfully benefits the company’s bottom line, as well as team morale. By incorporating a few small changes, the company can enable their workers to work how and when they want, without sacrificing productivity.</p><ul><li>Group meetings should be fully remote if one person is remote. No 10 people in a conference room and 1 person on Zoom. Everyone on the meeting should zoom into the meeting.</li><li>Document norms around time zones and when the team will respond to notifications. If Luke needs to pick up his daughter from school at 2pm PST every day, know that he’ll be back online later to respond.</li><li>Use remote as a competitive advantage. When hiring remotely, the world is your talent pool. Not only is the talent enormous outside of typical tech hubs, but Remote-first is a highly sought after benefit for knowledge workers.</li><li>Move away from the mindset that everyone needs to be at HQ. In fact, maybe get rid of your HQ — you’d probably save $$$$$$$$$ lots of money. IRL Offices are expensive and the costs saved can be spent on retention or a new hire.</li><li>When meeting in real life, focus on creating experiences for small teams within an org, to keep logistics and costs manageable.</li></ul><p><strong><em>Remote at scale: Company size, 1000+</em></strong></p><p>In general, a company seeking to transition to remote should actively avoid or remove common <a href="https://about.gitlab.com/company/culture/all-remote/hybrid-remote/#disadvantages-to-hybrid-remote">hybrid-remote pitfalls</a> by whatever means necessary. Moreover, it’s important for leaders to put themselves in the shoes of a prospective job seeker who will ask <a href="https://about.gitlab.com/company/culture/all-remote/evaluate/">very specific questions</a> to determine how prepared a company is to thrive in a remote setting.</p><p>Much of it boils down to changing whatever elements are needed in order to become the remote company that you would want to work for. This requires old habits to be left behind, and intentional effort to be made on areas such as <a href="https://about.gitlab.com/company/culture/all-remote/handbook-first-documentation/">documentation</a>, <a href="https://about.gitlab.com/company/culture/all-remote/informal-communication/">informal communication</a>, and working well <a href="https://about.gitlab.com/company/culture/all-remote/asynchronous/">asynchronously</a>.</p><p>Given the relative newness of all-remote, many people wonder if a 100% remote company works well at scale. This is largely due to unfamiliarity with the concept, and it overlooks a different question — does colocation work well at scale? (Narrator: “No… no, it does not.”)</p><p>GitLab believes that all-remote is the future of work, and that it not only works well at scale, but works <em>better</em> at scale than antiquated colocated models.</p><ul><li>The <em>quickest</em> way to send the <em>clearest</em> signal that remote is the future is to start at the top of the organizational chart. Remove execs from the office, and you’ll quickly figure out what gaps you need to fill with tools and process.</li><li>Maintain a team devoted to helping people acclimate to remote, answering questions about workspaces, workflows, conflict resolution, etc.</li><li>Reinforce values <em>daily</em>, not just once during onboarding. Consider implementing <a href="https://about.gitlab.com/handbook/incentives/#discretionary-bonuses">discretionary bonuses</a> that are directly tied to exemplifying values, and share each one granted with the entire company.</li><li>Scaling and transitioning are processes, and you should approach them with a spirit of patience and iteration. Expect bumps along the road. As with any significant business transformation, it’s wise to communicate proactively to team members, customers, and investors that obstacles will emerge. Transitioning a company to fully remote, while still running the business, is not easy. It’s a long-term bet that the short-term pain will be <a href="https://about.gitlab.com/company/culture/all-remote/benefits/">worthwhile</a>.</li><li>Work <a href="https://about.gitlab.com/company/culture/all-remote/handbook-first-documentation/">handbook-first</a>, always. Documentation isn’t enough — you need a universally understood single source of truth. If it’s not there, it doesn’t exist. This forces work to occur in the place where it needs to end up, rather than splintered across channels.</li><li><a href="https://about.gitlab.com/company/culture/all-remote/asynchronous/">Asynchronous communication</a> works best when there is companywide alignment on how and where to input communication. Leaders should carefully select their tools, aiming to direct communications to as few channels as possible. The easiest way to enter into an asynchronous mindset is to ask this question: “How would I deliver this message, present this work, or move this project forward right now if <strong>no one</strong> else on my team (or in my company) are awake?”</li><li>Above all, a company which intends to begin hiring remotely (or allowing existing employees to optionally transition from colocated to remote) must structure the company as if <a href="https://about.gitlab.com/company/culture/all-remote/transition/">every single team member were remote</a>. No shortcuts, no exceptions. It’s remote-first for everyone.</li></ul><p><strong>Remote for life</strong></p><p>We shouldn’t have to wait for public health crises to make changes in how we work for the better. In a world where we’re always connected and <strong>can </strong>work from anywhere, why don’t more of us make the conscious choice to actually do it?</p><figure><img alt="Working Remote in 2020" src="https://cdn-images-1.medium.com/max/560/1*UbfrIPcFZPLuRYelEV-dFg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=56122126f3df" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Growing Remote]]></title>
            <link>https://medium.com/@natemcguire/growing-remote-4f774df06fed?source=rss-9889a2a5666c------2</link>
            <guid isPermaLink="false">https://medium.com/p/4f774df06fed</guid>
            <category><![CDATA[allremote]]></category>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[hiring]]></category>
            <category><![CDATA[remote]]></category>
            <dc:creator><![CDATA[Nate McGuire]]></dc:creator>
            <pubDate>Fri, 07 Feb 2020 15:28:51 GMT</pubDate>
            <atom:updated>2020-02-07T15:51:11.361Z</atom:updated>
            <content:encoded><![CDATA[<h3>50 people in 17 countries and who knows how many time zones</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lNE4G5wEkCUD463dCM55BQ.png" /></figure><p><strong>Hi, I’m Nate.</strong></p><p>I’m one of the founding partners at <a href="https://mayvenstudios.com/">Mayven</a>, a 100% remote design and engineering agency with over 50 people, where we help Silicon Valley brands and startups scale. To support our team, we also built our agency’s operating system, <a href="https://buildstack.com/">Buildstack</a>, which helps remote teams run their business.</p><p>I’m usually in San Francisco, but can be frequently found on a plane, or visiting my co-founder Connor in his current country of choice.</p><p>Our passion is having the freedom to build and experiment with new products, people, and ideas, to help them grow. All while working and living remotely. I’ve lived and worked fully remote for nearly 10 years, and am excited for the future of fully remote teams and the software that helps make it all possible.</p><p><strong>Remote is sooooo hot right now, but why?</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/620/1*hXj3xG5giPZAIrFjudN8_g.jpeg" /></figure><p>VC Twitter is super into remote working right now. Maybe it’s the insane costs of living in the San Francisco Bay Area. Maybe it’s the millions of dollars companies waste every year on sweet offices. Or maybe they realize there are 7 billion smart people to recruit from that don’t live in California.</p><p>Remote work has quickly become one of the most desirable benefits for job seekers. They want the <strong>freedom</strong> to work from anywhere, with more flexible hours.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VoFmnqRYOqiAvCumyCupOQ.png" /><figcaption><a href="https://twitter.com/andrewchen/status/1074102659253133313">https://twitter.com/andrewchen/status/1074102659253133313</a></figcaption></figure><blockquote>Between 2012 and 2016, <a href="https://www.gallup.com/workplace/238085/state-american-workplace-report-2017.aspx">Gallup data</a> showed an increase in the number of employees who worked <strong>spent at least some of their time working remotely. </strong>The percentage of time spent working remotely increased as well.</blockquote><blockquote>I bet they wish it was 100% of the time.</blockquote><blockquote>According to <a href="https://globalworkplaceanalytics.com/">Global Workplace Analytics</a>, as of 2018 there are 4.3 million remote workers in the United States, which make up 3.2% of the entire workforce.</blockquote><p><strong>TINY BUT GROWING.</strong></p><p>It’s easy to see why we advocate for remote work:</p><ul><li>There are fewer distractions: Does anyone like open office floor plans? Except for maybe Jim, that guy who talks too loud. He loves it. Too much.</li><li>Recruit globally and hire the best talent</li><li>Reduces stress, especially due to commute</li><li>Flexible work schedule: Just get the job done</li><li>Save 408 hours per year by not traveling to work (<a href="https://www.theladders.com/career-advice/topic/latest-news">TheLadders.com</a>)</li><li>Total focus on what you produce, not on time spent</li><li>Have a life! Pick up your kids from school, take care of your elderly parents, go on that vacation. Doesn’t matter to us where you are.</li></ul><p>This is why Connor and I built a fully remote team — neither of us are interested in commuting and sitting in an office everyday and our team isn’t either. It doesn’t make sense why anyone would want to zap your team of productivity by making them commute.</p><p><strong>How can remote teams work better?</strong></p><ul><li>Communication is key in life, but even more crucial on a remote team. The better written communication your employees have, the better off you will be as a remote organization. It’s important to commit to this style of async work and not rely on video calls and massive amounts of meetings as a crutch.</li></ul><p><strong>What’s missing?</strong></p><ul><li>The tools and internet access are there. Now we need to create awareness of remote and how it favors the health of workers and their productivity over an old-school “butts in seats” mentality.</li></ul><p><strong>We’ve learned a lot, and I’m sure there’s more to come</strong></p><p>When we launched Mayven over five years ago, we just did everything using email, google docs, git, slack, and freshbooks. Which worked for the first few projects. Then, we went from about $150,000 in revenue in 4 months to over $1m in our first full year and knew we needed something a little more…robust.</p><p>We realized that to scale our efforts, maintain our freedom, and manage a team with the same ethos — work hard, asynchronously, and live your life how you want — we needed to have an integrated system that worked for us, as a remote team. There’s plenty of tools for non-remote teams, but an app built remote first for a remote services team to run the business.</p><p>We needed to be better at communicating and collaborating with clients in our never ending goal to alleviate <strong>project chaos. </strong>When you’re working with clients, it’s important for communication to be clear, documented and have tangible actions. When you work with a client remotely, you have to do those all 10x as well.</p><p>Over time, Buildstack came together as our system of record — it certainly helps us to look legit to clients, but more importantly gives our team a sense of <strong>place to work</strong>. When we tell our services customers that we created Buildstack, the most common reaction is “Wait really? WOW”. Instant credibility from the most fickle of engineers.</p><p>Some of the brands we’ve worked with as an all remote team: Airbnb, Andreessen Horowitz, First Round Capital, Google, Hired, Mixpanel, YouTube, Facebook, Instagram, IBM, TaxAct, Bolt, Homelight and more.</p><p>Our fully remote team started with just a few team members scattered across the Americas and Europe. Generally we’ve found sourcing from various contractor platforms to be the best bet, then using a test project (paid of course) to see if people are as good as they interview. This allows us to assess team members based on need, communication skills, and a “real world” scenario. We now are at 50 people in 17 countries and time zone mastery is second nature.</p><p>Many of our team members work as freelancers prior to joining us, so we have different expectations around the amount of hours someone should work, meaning you can be a meaningful contributor to the business at even 15, 30, or 50 hours a week if that’s how much you want to work. This gives us an advantage in recruiting senior talent who emphasize time with family, or their sanity and freedom, over putting in 100 hour work weeks in an office.</p><p>As we scale, we certainly continue to face challenges. The biggest always comes down to <strong>finding the right talent </strong>— not much different than IRL companies. Our advantage? We have a global talent pool to source from, making it possible to find and onboard new people super quickly and offer them freedom that an IRL job can’t.</p><p>Over time Connor and I have developed communication short-hand for how we work together. We like to joke we have the longest running chat threads in history since it’s just a constant stream of messages — mostly asynchronous. I also have his timezone permanently on my Apple Watch, and whether he’s living in Indonesia or Hong Kong, we talk every morning and night. Mayven ends up being a 24/7 magic making machine.</p><p><strong>Why Remote is the Future</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*UbfrIPcFZPLuRYelEV-dFg.jpeg" /><figcaption>Mayven Hong Kong</figcaption></figure><p>We continue to see remote as the future of work — so much that it will just become <strong>work</strong>, not “remote work”. In an economy primarily driven by services and increasingly knowledge based digital goods it simply is a better, more flexible way to work. If you live on your laptop, and drive an hour to work to sit in front of your laptop…why??? Of course not every job that is necessary and valuable will be done remotely, but in a world where many of us work digitally, there’s no reason to be in an office 9 to 5.</p><p>PS — I count 8 people at this coffee shop I’m at today working remotely. Shout out to Houndstooth in South Austin. Mayven was started the same way with nothing but a laptop and dash of hustle.</p><p><em>Speaking of Austin, I will be on two upcoming panels — One at SXSW in March </em><a href="https://schedule.sxsw.com/2020/events/PP1137477"><strong><em>“Modern Work: Finding Structure in an Officeless World”</em></strong></a><strong><em> </em></strong><em>and coming up quickly on February 26th, </em><a href="http://bit.ly/SafetyWingSFPrivate"><strong><em>“How to Build and Scale a Remote team”</em></strong></a><strong><em> </em></strong><em>— join us and continue the conversation.</em></p><p><strong>Nate McGuire</strong></p><p>Founder <a href="https://mayvenstudios.com/">Mayven</a> and <a href="https://buildstack.com/">Buildstack</a>.</p><p>Based in San Francisco, but I don’t always work there. I write about how teams can work and grow together while living remotely.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4f774df06fed" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>