<?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 Drew Tavernier on Medium]]></title>
        <description><![CDATA[Stories by Drew Tavernier on Medium]]></description>
        <link>https://medium.com/@drewtavernier?source=rss-1714908ae96a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*wp8v2Gh5-_FkfLs2ugBZ0g.png</url>
            <title>Stories by Drew Tavernier on Medium</title>
            <link>https://medium.com/@drewtavernier?source=rss-1714908ae96a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 10:41:35 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@drewtavernier/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[AI Is Making Experts Better. What Happens to Everyone Else?]]></title>
            <link>https://drewtavernier.medium.com/ai-is-making-experts-better-what-happens-to-everyone-else-08e07abd729d?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/08e07abd729d</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[critical-thinking]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Mon, 07 Sep 2026 06:06:00 GMT</pubDate>
            <atom:updated>2026-09-07T06:06:00.882Z</atom:updated>
            <content:encoded><![CDATA[<p>I’ve been thinking a lot about AI recently, and not in the usual “it’s going to replace everyone” kind of way.</p><p>Most of my job is thinking.</p><p>Whether I’m solving problems for clients, designing products, challenging assumptions or helping teams make better decisions, the work is rarely about having the answer. It’s about asking better questions, spotting what everyone else has missed and being comfortable sitting with uncertainty until something starts to make sense.</p><p>A big part of what I do is also teaching other people how to think that way.</p><figure><img alt="A whiteboard with post-its on" src="https://cdn-images-1.medium.com/max/1024/1*taO--s_-HNTWilxyMAJfUQ.png" /></figure><p>That’s why some of the AI content doing the rounds makes me uncomfortable.</p><p>You know the posts.</p><blockquote>“These three prompts will…”</blockquote><blockquote>“Copy this and become an expert.”</blockquote><blockquote>“Build a million-pound business with ChatGPT.”</blockquote><p>I understand why they do well. We all like the idea that there’s a shortcut, especially when the technology feels overwhelming.</p><p>The problem is that prompting isn’t the difficult bit, <strong>thinking is.</strong></p><p>Right now we’re in a strange place. The people getting the most out of AI tend to be the people who already know their subject inside out. They know when something doesn’t look right. They know what questions haven’t been asked. They can spot the assumptions hiding in an answer because they’ve spent years learning the subject before AI ever arrived.</p><p>AI is making experts better.</p><p>At the same time, we’re seeing people rely on AI without having that foundation. If you’ve never learnt how to evaluate an argument, question evidence or understand a problem properly, how do you know when AI has taken you in the wrong direction?</p><p>You probably don’t.</p><p>The tricky thing is that AI is incredibly convincing. It doesn’t just give you answers. It gives you answers with confidence. If you don’t have enough knowledge to challenge what it’s tells you, confidence can easily be mistaken for correctness.</p><p>That’s the bit I worry about.</p><p>Not because I think people are lazy, but because critical thinking is a skill. Like any skill, if you stop using it, it fades. If every difficult decision is handed over before you’ve wrestled with it yourself, you never build the judgement that tells you when something feels off.</p><p>Ironically, AI has made me think more, not less.</p><p>Most of my prompts aren’t asking for answers. They’re asking for resistance.</p><ul><li>Challenge this idea.</li><li>Tell me where my reasoning falls apart.</li><li>What assumptions am I making without realising it?</li><li>What edge cases have I ignored?</li><li>Play devil’s advocate until I’m convinced the idea survives.</li></ul><p>Those conversations often leave me with more questions than answers, and I’m completely fine with that because that’s usually where the interesting work starts.</p><p>I don’t think the future belongs to the people with the cleverest prompts.</p><p>I think it belongs to the people who can still think critically, who know when to trust AI, when to challenge it and, just as importantly, when to ignore it altogether.</p><p>Thanks for reading.</p><p>I write about the things I’m curious about and the things I’m building. Sometimes that’s service design, AI, product development and security. Other times it’s productivity, martial arts, meditation, or lessons from everyday life.</p><p>This is my little corner of the internet where I try to make sense of interesting ideas, share what I’m learning, and occasionally challenge my own assumptions along the way.</p><p>If you’d like to follow what I’m working on or explore some of the projects I’m building, you can find me here.</p><p>You can learn more at:</p><ul><li><a href="http://www.hueylabs.com/">HueyLabs</a></li><li><a href="https://drewtavernier.medium.com/">Medium</a></li><li><a href="https://www.linkedin.com/in/drew-tavernier/">LinkedIn</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=08e07abd729d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[One Prompt Isn’t Going to Make Your App Secure]]></title>
            <link>https://drewtavernier.medium.com/one-prompt-isnt-going-to-make-your-app-secure-ea76c83e2ab1?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/ea76c83e2ab1</guid>
            <category><![CDATA[data-security]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[ai-prompts]]></category>
            <category><![CDATA[application-architecture]]></category>
            <category><![CDATA[software-testing]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Mon, 24 Aug 2026 18:36:01 GMT</pubDate>
            <atom:updated>2026-08-24T18:36:01.361Z</atom:updated>
            <content:encoded><![CDATA[<p>Has anyone else’s feed been full of posts claiming you can secure your app with a single prompt, or save thousands of tokens by copying someone’s AI workflow?</p><p>I’ve seen them too.</p><figure><img alt="A prompt box with Make my app secure and underneath the reality of the complexity is takes to do that" src="https://cdn-images-1.medium.com/max/1024/1*veSwrwcZKtOkB9T6uVD-kQ.png" /><figcaption>Can teh magic prompt replace good engineering?</figcaption></figure><p>Over the last couple of months, it feels like every other post has been promising a shortcut to building software with AI. At the same time, I’ve been seeing more and more founders and indie hackers posting about apps getting hacked, API bills spiralling because rate limiting wasn’t implemented, authentication issues exposing user data, and production apps falling apart because the vibe coded version wasn’t ready for the real world.</p><p>AI has made building software more accessible than ever, and I genuinely think that’s a good thing. People who would never have written a line of code a year ago are now building products and shipping them.</p><p>I’m one of them. I’m not an engineer by trade.</p><p>What worries me is the growing narrative that engineering best practices have somehow become optional because AI can generate code quickly.</p><p>Good software has never just been about writing code. Security, architecture, testing, documentation, and operational decisions are all part of building something people can trust.</p><p>You can’t replace those things with a prompt.</p><p>A couple of months ago, I stopped bookmarking prompt libraries and started building something I actually wanted to use myself. A security first framework for AI assisted development.</p><p>The goal wasn’t to build another SaaS boilerplate or starter template. There are already plenty of excellent options out there. I wanted to build something that helps developers and non developers think about the things that rarely make it into viral LinkedIn posts.</p><p>Things like:</p><ul><li>Have you considered rate limiting before exposing an API endpoint?</li><li>Do you have a threat model for handling user data?</li><li>Is your authentication flow secure?</li><li>Are you testing what you’ve built?</li><li>Have you documented your architecture so your AI tools understand the project?</li><li>Are your LLM integrations protected against common security issues?</li></ul><p>When you’re not an engineer, it’s incredibly difficult to know what you don’t know.</p><figure><img alt="An icebery with top level what you think you need to build an app versus what you actually need" src="https://cdn-images-1.medium.com/max/1024/1*4Bdp18TigJ9LFyNRyuZUYA.png" /><figcaption>What you think you need, versus what you need</figcaption></figure><p>AI is great at helping you implement something, but unless you’ve established the right constraints and guardrails, it’ll happily build exactly what you’ve asked for, even if what you’ve asked for isn’t secure or production ready.</p><p>Starter Dev Kit doesn’t solve that problem entirely, and it isn’t intended to replace an engineering team. Experienced engineers are still invaluable when building serious products, and no framework or prompt is going to change that.</p><p>What it does provide is a much better starting point than opening Cursor and typing, “make my app secure.”</p><p>The framework bakes security and engineering practices into the development process from the beginning. It includes specialised AI agents, development rules, project scaffolds, testing requirements, security standards, and production checklists that encourage you to think about building software as a system rather than a collection of prompts.</p><p>The philosophy behind it is simple.</p><p>AI makes implementation cheap, which means we can afford to be more complete, not more careless.</p><p>I’m far less interested in saving a few tokens than I am in helping people avoid the kinds of mistakes that cost thousands of dollars, expose customer data, or force them to rebuild their product a month after launch.</p><p>AI assisted development isn’t going anywhere, and I don’t think it should. The tools are incredible, and they’re only getting better.</p><p>What I do think is missing from the conversation is a greater focus on engineering discipline and security.</p><p>Starter Dev Kit is my attempt to contribute to that conversation. It’s not a magic prompt, and it won’t replace good engineers. It’s simply a framework designed to help people build with AI more thoughtfully, with security and production readiness considered from day one.</p><p>Oh and I built a Claude Skill to help you write the docs to kick it off. Enjoy :)</p><p><a href="https://github.com/hueylabs/starter-dev-kit">GitHub - hueylabs/starter-dev-kit: starter security first dev kit</a></p><p>Thanks for reading.</p><p>I write about the things I’m curious about and the things I’m building. Sometimes that’s service design, AI, product development and security. Other times it’s productivity, martial arts, meditation, or lessons from everyday life.</p><p>This is my little corner of the internet where I try to make sense of interesting ideas, share what I’m learning, and occasionally challenge my own assumptions along the way.</p><p>If you’d like to follow what I’m working on or explore some of the projects I’m building, you can find me here.</p><p>You can learn more at:</p><ul><li><a href="http://www.hueylabs.com/">HueyLabs</a></li><li><a href="https://drewtavernier.medium.com/">Medium</a></li><li><a href="https://www.linkedin.com/in/drew-tavernier/">LinkedIn</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ea76c83e2ab1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[You Can’t Prompt Experience]]></title>
            <link>https://drewtavernier.medium.com/you-cant-prompt-experience-f4ed0c1c1fa3?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/f4ed0c1c1fa3</guid>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Mon, 10 Aug 2026 06:31:01 GMT</pubDate>
            <atom:updated>2026-08-10T06:31:01.432Z</atom:updated>
            <content:encoded><![CDATA[<p>AI is a new tool, but a carpenter doesn’t build a beautiful chair simply because they own a new chisel; they build it because of the years they’ve spent learning their craft.</p><p>The chisel matters, yet it’s the experience behind it that determines when to use it, how to use it, and what to do when things don’t quite go to plan.</p><p>AI is no different.</p><figure><img alt="A beautiful wooden handmade chair" src="https://cdn-images-1.medium.com/max/1024/1*cJms4ML1WWsLiLUeaFNlWw.png" /><figcaption>The master creates, not the chisel</figcaption></figure><p>Lately, I’ve been talking to a lot of people who are genuinely worried about where they fit in an AI-enabled world, and one conversation that stuck with me was with an amazing Business Analyst who feared that AI would make their skills less valuable.</p><p>My reaction was almost the opposite.</p><p>Your superpower isn’t knowing how to write the perfect prompt; it’s the years you’ve spent understanding people, processes, systems, risks and trade-offs. You’ve spent years helping organisations build complex things that actually work, and that kind of experience can’t be replicated in a prompt.</p><p>AI can help you move faster, but it can’t replace the judgement you’ve developed over years of doing the work.</p><p>I think we’ve become so focused on what AI can build that we’ve forgotten to ask who should be using it and why.</p><p>Product owners, Business Analysts, Service Designers and Product Designers don’t just build things; we understand edge cases, we know where systems break down, and we recognise that customers don’t experience products one screen at a time.</p><p>Take Service Ddesign as an example: AI can generate screens, workflows and user journeys in seconds, which is impressive, but customers don’t interact with services in isolation. They have goals, and they move <a href="https://drewtavernier.medium.com/customers-follow-journeys-organisations-follow-org-charts-9e3502d367d2">horizontally across organisations</a>, channels and touchpoints to achieve those goals.</p><p>And we don’t guess at that, we learn it through user research, conversations and observation, noticing where people struggle, where they drop off, what frustrates them and what delights them. We see the gaps between what was designed and what actually happens in the real world. That’s where the value sits.</p><figure><img alt="Notebook and screen showing a flow diagram" src="https://cdn-images-1.medium.com/max/1024/1*tfsp-i4ZNgm4Pd0ijZYbUA.png" /><figcaption>Using our experience to design systems</figcaption></figure><p><strong>A good service designer doesn’t just ask, “What should this screen look like?” but instead asks, “What is the customer trying to achieve, and how does every part of this service help them get there?”</strong></p><p>AI doesn’t know your customers, nor does it understand organisational context, competing priorities or the thousand small details that make an experience delightful rather than frustrating, which is why it needs people with experience to guide it.</p><p>That’s why I think it’s time we stop viewing AI as something that replaces expertise and start seeing it for what it really is: a force multiplier.</p><p>Bad decisions still create bad products, no matter how good the tools are, and while AI can amplify brilliant thinking, it can just as easily amplify poor thinking if we aren’t intentional about how we use it.</p><p>In many ways, AI has removed the entry barrier for building things, meaning more people can build applications, prototypes and products than ever before, which is a good thing.</p><p>But building something has never been the difficult part; building the right thing is.</p><p>That’s where experience comes in, because we ask the questions AI can’t answer on its own: should we build this, who does it help, does it deliver value, is it solving the right problem, and are we making life better for the people using it?</p><p>Those questions come from years of experience, mistakes, conversations and learning, and they are deeply human.</p><p>There is one thing I do worry about, though.</p><p>AI makes it easier than ever to jump straight to the end result, which is exciting, but what happens if we start skipping the experience phase altogether?</p><p>Future generations will grow up with AI as the default, and while they’ll be able to build faster than any generation before them, which is powerful, if we overvalue outcomes we risk undervaluing the journey that creates real expertise.</p><p>Because expertise isn’t something you download; it’s something you earn.</p><p>It’s built through mistakes, frustrations, bad decisions and those moments where you think you’ve solved the problem only to realise you’ve created three more, and most of the lessons that matter don’t come from getting it right but from understanding why it went wrong.</p><p>There’s no perfect path, only progress, and progress only happens when you walk it.</p><p>One of the ideas I love from Taoist philosophy is that the Tao, or “the Way”, isn’t somewhere you arrive; the path itself is the point, and you become who you are by walking it.</p><p>That resonates deeply when I think about AI.</p><figure><img alt="Kintsugi bowl on a wooden table" src="https://cdn-images-1.medium.com/max/1024/1*1L_nODMfZJT56NZcz1nf6w.png" /><figcaption>We learn from imperfection</figcaption></figure><p>You don’t become a great Business Analyst because AI writes better requirements; you become one because you’ve spent years understanding complexity and helping people make sense of it.</p><p>You don’t become a great Service Designer because AI generated a journey map; you become one because you’ve sat with customers, challenged assumptions and learned that services are messy because people are messy.</p><p>You don’t become a great product owner because AI helped you write a backlog; you become one because you’ve learned what happens when you build the wrong thing, even when you’ve built it exceptionally well.</p><p>The most valuable lessons are rarely at the start or the finish; they’re found in the middle.</p><p>AI can help us move faster, but it can’t walk the path for us.</p><p>If we’re not careful, we’ll optimise for speed and forget that some things are valuable precisely because they take time to learn, because mastery isn’t about reaching a finish line but about continuously refining how we think, how we build and how we serve others.</p><p>That’s the real opportunity with AI: not to replace the journey, but to remove enough of the mundane that we can spend more time on the parts that make us better.</p><p><strong>My take is simple: human expertise is the foundation, and AI is the tool that amplifies it when used intentionally and well.</strong></p><p>The future isn’t AI versus humans; it’s experienced humans using AI to build better products, better services and better outcomes.</p><p>The tool may change, but the path doesn’t.</p><p>And ultimately, we don’t just build things, we decide what deserves to be built at all.</p><p>Thanks for reading.</p><p>I write about the things I’m curious about and the things I’m building. Sometimes that’s service design, AI, product development and security. Other times it’s productivity, martial arts, meditation, or lessons from everyday life.</p><p>This is my little corner of the internet where I try to make sense of interesting ideas, share what I’m learning, and occasionally challenge my own assumptions along the way.</p><p>If you’d like to follow what I’m working on or explore some of the projects I’m building, you can find me here.</p><p>You can learn more at:</p><ul><li><a href="http://www.hueylabs.com/">HueyLabs</a></li><li><a href="https://drewtavernier.medium.com/">Medium</a></li><li><a href="https://www.linkedin.com/in/drew-tavernier/">LinkedIn</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f4ed0c1c1fa3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Customers Follow Journeys. Organisations Follow Org Charts.]]></title>
            <link>https://drewtavernier.medium.com/customers-follow-journeys-organisations-follow-org-charts-9e3502d367d2?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/9e3502d367d2</guid>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Mon, 22 Jun 2026 12:26:00 GMT</pubDate>
            <atom:updated>2026-07-23T17:43:33.490Z</atom:updated>
            <content:encoded><![CDATA[<p>Customer journeys and service blueprints are often presented as the answer to organisational complexity.</p><p>I don’t think that’s quite right. To me, they’re tools that help us understand complexity.</p><p>Most organisations are built vertically. Teams own products, platforms, services, features, systems, pages, and roadmaps. That structure exists for good reason. People develop expertise in specific domains, build context over time, and learn how to deliver well within their part of the world.</p><p>Ask a team to suddenly work in an unfamiliar area and things usually slow down. New questions appear. New dependencies emerge. Eventually, somebody else inherits the problems that were accidentally created along the way.</p><p>So vertical ownership isn’t the problem. The challenge is that customers don’t experience organisations vertically.</p><p>Customers don’t wake up wondering which team owns authentication, who manages fulfilment, or which platform is responsible for notifications. They have a goal, and they move horizontally through the organisation trying to achieve it.</p><p>Along the way, they cross teams, systems, channels, policies, and processes, carving a path through the organisation like water finding its way through a mountain. Over time, that flow becomes a river.</p><p>Service design helps us see it.</p><figure><img alt="Customer Journey flowing through an org chart" src="https://cdn-images-1.medium.com/max/1024/1*lG0gCmjkR1e41Pkt98BhAg.png" /></figure><p>Journey maps help us understand what customers are trying to achieve, how they think, what they feel, where they hesitate, and where frustration starts to creep in. Service blueprints help connect those experiences to the machinery behind the scenes: the systems, teams, processes, policies, dependencies, and business decisions that shape what customers ultimately experience.</p><p>Together, they give us an aerial view of the landscape. Without them, we’re often stuck in street view, seeing only the small section directly in front of us.</p><p>But nobody ever fixed a customer problem because they produced a beautiful journey map, and no customer has ever said, “Thank goodness the stickies were colour coordinated.”</p><p>The map isn’t the value, and neither is the blueprint. The value comes from creating them.</p><p>The real work happens in the conversations that emerge while they’re being built. It happens when Product, Engineering, Operations, Customer Support, Commercial teams, and Designers sit together and attempt to describe the same journey.</p><p>What starts as a straightforward workshop usually becomes something much more interesting.</p><ul><li>Someone discovers a workaround they’ve quietly accepted for years.</li><li>Someone realises a policy decision created friction they never intended.</li><li>Someone learns customers are doing far more work than anyone thought.</li><li>Someone discovers the thing they believed belonged to another team actually belongs to nobody at all.</li></ul><p>Those moments are where change begins, not because a diagram exists, but because understanding exists.</p><p>That’s why the most valuable part of service blueprinting is often the shared understanding that emerges through the process.</p><p>When we bring stakeholders into the process, we’re not just documenting a service. We’re helping people see it differently. We’re connecting business decisions to customer outcomes and exposing the invisible chain of events between a customer pressing a button and a product arriving at their door.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JE_VMUUf9JOdLY8_qUogdg.png" /></figure><p>We’re also exploring what happens when things don’t go to plan: when a delivery is delayed, when someone wants a refund, when a customer is confused, when they need help, or when they leave a one-star review and somebody has to figure out what happened.</p><p>The interesting questions rarely live on the happy path because customers don’t spend much time there.</p><p>One of the biggest risks of working on a service for a long time is that the workarounds become invisible. We adapt to the quirks, learn the shortcuts, and stop noticing the strange language, awkward handoffs, and unnecessary steps because they’ve become part of everyday life.</p><p>Customers experience these issues firsthand. Some manage to work around them, while others become frustrated and give up altogether.</p><p>For them, these problems are new. They’re often just trying to complete a simple task, which is a surprisingly effective way to turn an ordinary afternoon into a frustrating one.</p><p>Service design gives us a structured way to rediscover that reality. Not through certainty or assumptions, but through curiosity and better questions.</p><p>Where does the customer slow down? Where are we asking them to understand our organisation instead of solving their problem? Where are individual teams optimising their part of the experience while the overall journey becomes harder? Where have business decisions unintentionally increased effort for customers?</p><p>Those questions are important because they help us bridge two worlds: the world of organisational delivery and the world of customer experience.</p><p>That’s the role of service design.</p><p>Creating understanding.</p><p>The journey map on the wall isn’t the outcome, and the blueprint isn’t the outcome either. The outcome is a room full of people who can suddenly see the service through the customer’s eyes and make better decisions because of it.</p><p>It’s aligning business goals with customer needs instead of treating them as competing forces. It’s finding the gaps between teams, processes, and systems where frustration lives, then deciding to do something about them.</p><p>Life is messy. Services are messy. Organisations are definitely messy.</p><p>Trying to eliminate that mess entirely is probably a losing battle. Understanding it, though, is where the opportunity lies.</p><p>Because wherever things stretch, break, or create friction, there’s usually something worth learning. And if we’re willing to follow the river instead of the org chart, we might just find something worth fixing.</p><p>Thanks for reading.</p><p>I write about the things I’m curious about and the things I’m building. Sometimes that’s service design, AI, product development and security. Other times it’s productivity, martial arts, meditation, or lessons from everyday life.</p><p>This is my little corner of the internet where I try to make sense of interesting ideas, share what I’m learning, and occasionally challenge my own assumptions along the way.</p><p>If you’d like to follow what I’m working on or explore some of the projects I’m building, you can find me here.</p><p>You can learn more at:</p><ul><li><a href="http://www.hueylabs.com/">HueyLabs</a></li><li><a href="https://drewtavernier.medium.com/">Medium</a></li><li><a href="https://www.linkedin.com/in/drew-tavernier/">LinkedIn</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9e3502d367d2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI + ADHD… Acceleration or car crash]]></title>
            <link>https://drewtavernier.medium.com/ai-adhd-acceleration-or-car-crash-2c19378b5920?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/2c19378b5920</guid>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[self-improvement]]></category>
            <category><![CDATA[adhd]]></category>
            <category><![CDATA[creativity]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Thu, 04 Jun 2026 08:50:19 GMT</pubDate>
            <atom:updated>2026-07-23T17:43:51.913Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>AI didn’t fix my ADHD. It turbocharged it. And I’m still figuring out if that’s a good thing.</strong></p><p>I’m writing this between sets at the gym. Lat pulldowns, if anyone cares. Which feels weirdly appropriate, because this is how my brain works: half exercising, half writing, half planning another product idea. Yes, I know that’s three halves. Welcome to ADHD maths.</p><figure><img alt="3 brains working together" src="https://cdn-images-1.medium.com/max/1024/1*LGqp43-i7DDhYQe7IBQneA.png" /><figcaption>3 brains working together</figcaption></figure><p>I’ve always had a deeply curious mind. Not casually curious. Restlessly curious. My friends joke that I know 15% of everything, which is honestly fair. I’m the kind of person who can get dropped into almost any situation and figure something out. Think MacGyver, but with a MacBook and too many browser tabs.</p><p>When other people panic, I switch into hyperdrive. I can hold ten threads at once and usually find the connection nobody else spotted; the user need, the technical risk, the business model, the edge case, the flaw nobody wants to say out loud. <strong>I’ve come to see ADHD as a superpower.</strong> It has side effects, I can’t always tell you what I did five minutes ago, but mostly? It’s awesome.</p><p>The problem was never the thinking. It was everything that happened after.</p><figure><img alt="ADHD Superhero" src="https://cdn-images-1.medium.com/max/1024/1*ga_3P9cN_GcYKXSdLWzedw.png" /><figcaption>ADHD Superhero</figcaption></figure><p><strong>For most of my life, my ideas moved faster than my abilities. </strong>Not because I lacked ambition. Not because I was lazy. Because there was always a wall somewhere in the middle.</p><p>I’d get obsessed with an idea, a product, a tool, a system. I could see the finished thing clearly in my head. Then reality would arrive. An API I didn’t understand. Infrastructure problems. Authentication. Deployment. Some invisible engineering problem I didn’t know existed until I hit it face first.</p><p>So I’d do what a lot of people with ADHD do; obsessively learn. For two weeks I’d consume tutorials, documentation, YouTube videos, random forum posts from 2011 written by someone called Dave. Eventually I’d break through the wall. But by then the emotional momentum had usually disappeared and a newer idea had already arrived. So the cycle repeated.</p><p>For years I thought my problem was execution. Now I’m not so sure.</p><p><strong>AI changed the equation completely. </strong>The gap between imagination and execution has basically closed, and I don’t think most people have fully clocked what that means yet.</p><p>Right now I’m building three things at once. A travel site for Dominica, <a href="https://nature-island.com/">nature-island.com</a> , that I originally built in WordPress because there wasn’t a decent one, and I’m half Dominican. Left it alone for 18 months, checked Google Analytics on a whim, 200 visitors a day… but the site was clunky and Wordpress slows stuff down, so I rewrote the whole thing in SvelteKit &amp; Node.js, added my own content from a trip I took in October, drone shots, photography, actual experience of the island and I’m now building a feature that generates personalised day-by-day driving itineraries based on destinations people have favourited. A year ago that’s a few of months of work, now it’s taken me a month to rewrite all the content, with the help of Claude, and build a site that’s speedy like the wind.</p><figure><img alt="Destination page on nature-island.com" src="https://cdn-images-1.medium.com/max/1024/1*eACKUpTwKhxdczQ8lJWpIA.jpeg" /><figcaption>Destination page on nature-island.com</figcaption></figure><p>I’m also rebuilding WhizVid, my video production and scheduling tool, and planning a story-writing tool after I helped a friend structure a prompt system so he could finally finish his novel. Three projects. Simultaneously. They all feel optimal. I don’t. That’s the bit I’m still trying to understand.</p><p><strong>The strange thing about removing friction is that you find out what the friction was doing for you. </strong>Technical walls used to force prioritisation. You couldn’t build everything, so you had to choose. Now almost every idea feels actionable, immediately, and that’s dangerous for an ADHD brain.</p><p>Before AI, hard implementation phases acted like checkpoints. They made you stop and ask: do I actually care about this enough to keep going? A lot of the time the answer was no, and that was useful information. Now that question gets bypassed with another prompt. AI has turned every passing thought into a potential startup, every annoyance into a tool, every curiosity into a rabbit hole with instant momentum. ADHD brains already crave novelty, stimulation, and immediacy. AI delivers all three on demand. The dopamine loop gets almost frictionless.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m_5XinEv25bOi1pX8OfLKA.jpeg" /><figcaption>AI + ADHD</figcaption></figure><p>I got bad tendonitis a few months ago and couldn’t train. The most comfortable place was my office chair, Herman Miller, worth every penny, so I just stayed there. Eight hours of day job, then coding until I was tired enough to sleep, waking every four hours for pain meds. In that window I built an entire app. I wrote about it <a href="https://engineering.dunelm.com/if-you-can-think-it-you-can-build-it-badly-2bed8892db59">here</a>. Most people with an injury watch Netflix. I coded.</p><p>My wife had just launched her own business, <a href="https://www.givemeclarity.com/">Clarity Ahead</a>. We were both deep in our own things, two people in the same house, both building, both buzzing. Honestly it worked perfectly. At the time I told myself it was resourcefulness. Maybe it was. But it also showed me something; given zero friction and nowhere to be, I just build. That’s not a complaint. That’s just what I am.</p><p><strong>AI doesn’t just increase output. It amplifies whatever’s already inside you.</strong> Naturally focused? AI makes you more effective. Naturally scattered? AI makes you infinitely scatterable. The superpower gets louder.</p><p>If I’m not working on three things at once I feel unproductive, not logically, emotionally. There’s a low hum of guilt when I’m not building something. My new blockers aren’t technical, they’re biological and social. Sleep. Energy. Attention. The person in the same house who deserves more than a version of me with one eye on a code editor. Professionally, AI has been incredible. Genuinely. I’m building things I could only imagine two years ago. Psychologically, I’m less sure.</p><p>There’s one thing that makes it stop.</p><p>I’ve trained in martial arts most of my life. It’s always been the thing that keeps my brain genuinely occupied, not distracted, occupied. I had a stretch in the middle where work and projects filled that space instead, and looking back I was just substituting one intensity for another. When I came back to training I came back harder. For the past three and a half years that’s been Wing Chun, three times a week, more when I can find people to train with outside of class. I train to get better, faster, more focussed. That’s the whole point.</p><p>But something happens as a side effect. When I’m actually in it, not warming up, not drilling, but in the flow of it, everything goes quiet. Not quieter. Quiet. No product ideas, no architecture decisions, no background thread running through feature lists. Just the next half-second. It’s the only place I’ve found where the speed of thought isn’t an asset, where being two steps ahead gets you hit, where presence isn’t something you choose, it’s something a live opponent enforces. AI can accelerate almost everything I do. It cannot do that.</p><p>And right now, with the tendonitis still lingering, training is on hold. Circuit breaker gone. So the building continues, 6am to 8:45am before the day job starts, another hour after work because my wife doesn’t finish work until half six. The gaps in the day have become project time.</p><p>I spent years thinking; if only I could execute as fast as I think. Now I can. Turns out the speed was never the problem.</p><p>The walls were doing a job I didn’t know I needed done. Dave from the 2011 forum post was doing a job. Hitting a problem I couldn’t solve in an afternoon forced me to ask whether I actually cared. A lot of the time I didn’t. Useful information. Now nothing asks that question for me.</p><p><strong>AI gave people with ADHD something we always wanted, the ability to move at the speed of thought, the superpower fully unleashed.</strong> But maybe what we also needed, what we always needed, was the thing that made us stop. For me that’s a live opponent who hits back. I’m still figuring out what to do when they’re not there.</p><p><em>I’m building </em><a href="https://nature-island.com/"><em>nature-island.com</em></a><em>, a travel guide to Dominica built on personal experience. If you’ve ever thought about going, it’s one of the most extraordinary places on earth.</em></p><p>Thanks for reading.</p><p>I write about the things I’m curious about and the things I’m building. Sometimes that’s service design, AI, product development and security. Other times it’s productivity, martial arts, meditation, or lessons from everyday life.</p><p>This is my little corner of the internet where I try to make sense of interesting ideas, share what I’m learning, and occasionally challenge my own assumptions along the way.</p><p>If you’d like to follow what I’m working on or explore some of the projects I’m building, you can find me here.</p><p>You can learn more at:</p><ul><li><a href="http://www.hueylabs.com/">HueyLabs</a></li><li><a href="https://drewtavernier.medium.com/">Medium</a></li><li><a href="https://www.linkedin.com/in/drew-tavernier/">LinkedIn</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2c19378b5920" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Character Consistency Using MidJourney and Magnific]]></title>
            <link>https://engineering.dunelm.com/character-consistency-using-midjourney-and-magnific-601751e60b6c?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/601751e60b6c</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[character-design]]></category>
            <category><![CDATA[generative-ai]]></category>
            <category><![CDATA[creative-process]]></category>
            <category><![CDATA[midjourney]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Tue, 26 May 2026 07:01:00 GMT</pubDate>
            <atom:updated>2026-07-23T17:44:18.148Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>If you’ve ever tried to get character consistency in your designs, you’ve probably reached the point where you want to throw your laptop out the window, move to a desert island, and give up on AI forever.</strong></p><p>I completely understand.</p><p>I’ve had this idea for an animated series called <em>The Adventures of Frank &amp; Sparkle</em>. Frank and Sparkle are my mostly adorable pugs.</p><p>If you’ve ever owned pugs, you’ll understand the “mostly” part.</p><p>Strong-willed and stubborn are two of their many wonderful traits. Along with snoring like small broken tractors and looking deeply offended when you ask them to do anything reasonable.</p><p>Anyway, I digress.</p><p>I’ve been struggling with character consistency in MidJourney for a while. I could get close, but never quite close enough. One image would look great, then the next one would have a slightly different face, odd proportions, or the general haunted energy of “this is technically the same character, but also absolutely not.”</p><figure><img alt="First charater sheet attempt for Frank the pug" src="https://cdn-images-1.medium.com/max/1024/1*uwQsg74GnBVz-xjbJhAa3Q.png" /><figcaption>Initial attempt at a character sheet</figcaption></figure><p>The frustrating part is that each image might be good on its own. Sometimes really good. But when you put them next to each other, the differences become painfully obvious. The ears change. The face shape shifts. The outfit decides it wants a redesign. Suddenly your character has gone from “recurring cast member” to “artist’s interpretation loosely inspired by the original brief.”</p><p>That’s fine if you’re creating one-off images.</p><p>It’s a problem if you’re trying to build a world.</p><p>For anything story-based, consistency matters, the audience needs to recognise the character instantly, even if the pose, lighting, background, or expression changes. That’s what makes the whole thing feel intentional rather than like a collection of nice but unrelated images.</p><p>Then I found a technique by Gilbatree, which gave me the basis for creating really strong character sheets.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F5ISpGjtDijk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D5ISpGjtDijk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F5ISpGjtDijk%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/fb0a8837cb09d534ec90e23adcd8c17e/href">https://medium.com/media/fb0a8837cb09d534ec90e23adcd8c17e/href</a></iframe><h3>TLDR;</h3><p><strong>The basic steps for creating a good character sheet are:</strong></p><ol><li>Create your initial character on a white background, make sure it’s full body and you’re 100% happy with it as this will be the base of your character sheet.</li><li>Animate the character using your image as the start frame. You will create a lot of videos here, but there will be gold in each one.</li><li>Give it various prompts to create different poses and angles. Gilbatree explains this really well in the video and has resources to create them.</li><li>Download the videos.</li><li>Open them in QuickTime.</li><li>Right click and save the best frames as images.</li><li>Combine the images into 1 character sheet using your favourite image editor.</li></ol><p><strong>I think I went a bit overboard because I was having so much fun</strong></p><figure><img alt="Frank &amp; Sparkle Character Sheets" src="https://cdn-images-1.medium.com/max/800/1*IhJWOqH7FhYyaP5nXz8W4Q.png" /><figcaption>Frank &amp; Sparkle Character Sheets</figcaption></figure><p>The clever part of this process is that animation gives you lots of usable variations from the same starting point. Instead of asking the image model to reinvent the character every time, you’re using movement to generate different poses, angles, and expressions that still feel connected to the original design.</p><p>That character sheet then becomes a much stronger reference point. It gives the AI more information to work with than a single image ever could. Front view, side angles, expressions, body shape, proportions, little details, all the things that tend to quietly wander off when you’re only using one reference image.</p><p>Once you have a character sheet, you can continue using MidJourney as shown in the video.</p><p>That said, I couldn’t get it working quite as well as I wanted, especially when I had multiple characters in the same scene. One character would behave, the other would quietly turn into a distant cousin. AI likes to keep you humble.</p><p>This is where things started to get a bit more interesting.</p><p>I moved over to Magnific, formally Freepik.</p><p>I’ve only just started experimenting with it, but the initial results have been really promising.</p><p>I created a character reference group that included the two full-body characters side by side, so the scaling stayed consistent. I also added the full character sheets for both Frank and Sparkle.</p><p>That combination seemed to help a lot. The side-by-side reference gave the model a clearer understanding of how the two characters relate to each other physically, while the individual character sheets helped preserve the smaller details. It wasn’t perfect, but it felt much more controlled.</p><p>Then I added a scene reference image that I’d created in MidJourney and used Magnific to combine everything.</p><p>This part was important because I didn’t want to lose the look and feel of the environments I’d already built. MidJourney was still great for creating the overall scene style, mood, and composition. Magnific then became more of the tool for bringing the characters into that world with better consistency.</p><p>At least, that’s the working theory.</p><figure><img alt="Example flow in Magnific" src="https://cdn-images-1.medium.com/max/1024/1*gDBnXosgHyFBVxvOBvT6CQ.png" /><figcaption>Example flow in Magnific</figcaption></figure><p>I’m still refining the process, and there’s definitely more testing to do. I need to push it through more scenes, more poses, different lighting, and probably a few deeply unreasonable requests just to see where it breaks.</p><p>But as an initial workflow, I’m really happy with the results.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ffOEWn_2emC02lCqf8oNyQ.png" /><figcaption>Frank &amp; Sparkle in a magical forest</figcaption></figure><p>It feels like I’m finally getting closer to something usable, rather than spending hours arguing with a prompt box and pretending that “close enough” is fine.</p><p>The big lesson so far is that character consistency doesn’t seem to come from one magic prompt. Annoyingly, it’s more of a system. You need strong reference images, a good character sheet, clear scale relationships, and a workflow that gives the AI less room to improvise.</p><p>Which is slightly less glamorous than “just type the right words.”</p><p>But it works better.</p><p>And for AI character consistency, that feels like progress.</p><p>Thanks for reading.</p><p>I write about the things I’m curious about and the things I’m building. Sometimes that’s service design, AI, product development and security. Other times it’s productivity, martial arts, meditation, or lessons from everyday life.</p><p>This is my little corner of the internet where I try to make sense of interesting ideas, share what I’m learning, and occasionally challenge my own assumptions along the way.</p><p>If you’d like to follow what I’m working on or explore some of the projects I’m building, you can find me here.</p><p>You can learn more at:</p><ul><li><a href="http://www.hueylabs.com/">HueyLabs</a></li><li><a href="https://drewtavernier.medium.com/">Medium</a></li><li><a href="https://www.linkedin.com/in/drew-tavernier/">LinkedIn</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=601751e60b6c" width="1" height="1" alt=""><hr><p><a href="https://engineering.dunelm.com/character-consistency-using-midjourney-and-magnific-601751e60b6c">Character Consistency Using MidJourney and Magnific</a> was originally published in <a href="https://engineering.dunelm.com/">Dunelm Technology</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Remembering The Suffering: five years, 15 events, and 35,000 muddy finishers]]></title>
            <link>https://drewtavernier.medium.com/remembering-the-suffering-five-years-15-events-and-35-000-muddy-finishers-5cee9a2c0f03?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/5cee9a2c0f03</guid>
            <category><![CDATA[founders]]></category>
            <category><![CDATA[service-design]]></category>
            <category><![CDATA[obstacle-course-racing]]></category>
            <category><![CDATA[product-design]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Fri, 15 May 2026 06:14:37 GMT</pubDate>
            <atom:updated>2026-07-23T17:44:34.924Z</atom:updated>
            <content:encoded><![CDATA[<p>It’s amazing what your brain comes up with on a run.</p><p>Some people use that time to think clearly. They solve work problems. Plan dinner. Make sensible life choices.</p><p>I apparently decided the world needed an obstacle race called <strong>The Suffering</strong>.</p><p>In hindsight, that wasn’t the easiest thing to sell.</p><blockquote>“Fancy paying us money to suffer in a field?”</blockquote><p>Not exactly a warm invitation.</p><p>A fluffy bunny race would probably have been easier. Lure people in with something cute, then quietly send them up a hill carrying a sandbag while someone in a skull T-shirt tells them to do burpees.</p><p>But in 2012, after running a few obstacle course races that didn’t feel quite challenging enough, we put on our first event. That idea turned into five years of races, 15 events, and more than 35,000 runners crossing our finish line. We became <strong>an award-winning event</strong> and, in 2016, our Relentless Suffering Legends Race won UK’s Toughest Event, beating Tough Mudder and Spartan Race. We also won best swag that year for our finisher T-shirts, medals, and goody bags.</p><p>Not bad for a questionable idea on a run.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bRQVQ6OVcGRpt3Ui" /><figcaption>Our award winning swag</figcaption></figure><h3>Building the brand</h3><p>The obstacle racing market was full of events claiming to be tough.</p><p>We didn’t want to just say it. We wanted people to feel it by the finish line.</p><p>I designed the branding, the website, the user experience, the course routes, and the obstacles. So in a very real sense, I was responsible for both the customer journey and the part where that journey included mud, hills, water, crawling, climbing, and mild regret.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*zoAyWj2fJMbh_U_UunJGZA.jpeg" /></figure><p>The brand had to work before anyone got to the start line. The website needed to make the races feel exciting, intimidating, and achievable. The UX had to help people understand what they were signing up for, choose the right distance, trust us enough to buy a ticket, and then feel like they’d joined something bigger than a race.</p><p>That became a huge part of The Suffering.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Wx7uQKXwpyfLUzb3" /><figcaption>Our Race Team At The End Of The Suffering 10km Race</figcaption></figure><p>It wasn’t just a skull logo, although the skull logo did do a lot of heavy lifting. It was the race names, the tone of voice, the social posts, the emails, the medals, the T-shirts, the finish line, and every little interaction people had with us.</p><p>Our strapline was <strong>Be Your Own Hero</strong>.</p><p>A bit cheesy? Maybe.</p><p>But it worked because the event gave people a real chance to prove something to themselves.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*E7fixmpgYogg6He3" /><figcaption>Spideman Running The Suffering</figcaption></figure><h3>Designing the suffering</h3><p>My main role was designing the experience from screen to finish line.</p><p>I planned every inch of the courses. That sounds like an exaggeration, but it really wasn’t. I designed the route, the obstacle flow, the difficulty curve, the moments where runners could recover, and the moments where they’d quietly start questioning my character.</p><p>I worked out how to use the natural terrain properly. Streams, hills, woods, awkward climbs, muddy sections, open fields — all the bits that look harmless on a site visit and then become deeply personal on race day.</p><p>On big weekends, we had three courses totalling over 25 miles, with more than 30 obstacles ranging from 30-metre inflatable water slides to 20-metre rigs. I designed obstacles, tested ideas, wrote risk assessments, worked out working limits based on material strengths, and made sure the course could be brutal without being careless.</p><p>That balance mattered.</p><p>We wanted people to feel pushed, not unsafe.</p><p>Broken in spirit, perhaps.</p><p>But only temporarily.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F2FHFb7gt8dw%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D2FHFb7gt8dw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F2FHFb7gt8dw%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/2365017d4f8fa615127762cf574a9bd6/href">https://medium.com/media/2365017d4f8fa615127762cf574a9bd6/href</a></iframe><h3>The Reaper Squad</h3><p>The Reaper Squad became one of the best parts of The Suffering.</p><p>They were our race marshals, but with personality. Some were kind. Some were not. You might get jelly babies. You might get 20 burpees.</p><p>You never really knew.</p><p>For the runners, it felt chaotic and unpredictable. Behind the scenes, it was tightly managed.</p><p>We built a Reaper portal for volunteers with health and safety guidance, role expectations, conduct rules, and instructions on how to be a Reaper without accidentally turning the event into a lawsuit with mud on it.</p><p>Because you can’t just give someone a skull T-shirt and a radio, point them at tired runners, and hope for the best.</p><p>Well, you can.</p><p>But you shouldn’t.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*beFFuQ5-UTQyAagM" /><figcaption>A Reaper dressed as a reaper</figcaption></figure><p>We had a volunteer coordinator who looked after the Reapers properly, and around 20 paid, experienced race crew behind the scenes keeping the event moving. Every Reaper had an on-the-day briefing, a uniform — including the coveted Reaper T-shirt — and a radio back to base. That radio was the control mechanism. It kept the course coordinated and meant we could respond quickly if someone needed help, something changed, or a Reaper got a little too enthusiastic with the lunges.</p><p>Which did happen.</p><h3>Making people come back for more</h3><p>We did a few things that messed with people’s heads a bit.</p><p>In a nice way.</p><p>Mostly.</p><p>Every race distance was around 10% longer than advertised. We added extra obstacles. We used the landscape to make the course harder. We let the Reapers create moments of support or punishment.</p><p>It made the race physically hard, but mentally harder.</p><p>By the time someone finished, they didn’t just feel like they’d completed a race. They felt like they’d conquered something.</p><p>That mattered.</p><p>It also meant people came back. By year two, at least 40% of previous customers booked another event. By year three, that grew to over 50%.</p><p>Which is either loyalty or a worrying lack of self-preservation.</p><p>Possibly both.</p><h3>The messy bit</h3><p>People see the start line, medals, smoke, photos, mud, and smiles.</p><p>They don’t see the spreadsheet panic.</p><p>They don’t see the late-night course changes, stock deadlines, weather reports, risk assessments, volunteer briefings, signage problems, supplier calls, or the moment you realise an obstacle that looked perfectly reasonable in your head now needs six people, two ratchet straps, and a quiet prayer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*m98oCz7_-N1vWB4i" /><figcaption>A runner wading through a small stream</figcaption></figure><p>The 2016 floods were probably the purest version of The Suffering we ever had. Setting up a 10-mile course in those conditions was ridiculous. Wet, cold, muddy, exhausting.</p><p>Also, weirdly, brilliant.</p><p>That’s the bit I remember most fondly. Not because it was easy, but because everyone pulled together.</p><p>A damp joy.</p><p>But joy.</p><h3>The end</h3><p>In 2017, Dirty Dozen OCR, one of our main competitors, went out of business. That changed customer behaviour almost overnight.</p><p>Before that, people bought race tickets six months in advance. Afterwards, they started waiting until four to six weeks before race day because they were worried events might get cancelled.</p><p>Fair enough.</p><p>But for an event business, cash flow is oxygen. When advance sales slow down, everything gets harder. Medals, T-shirts, staffing, infrastructure, planning. All of it.</p><p>Eventually, we closed the doors in 2017 after five incredible years.</p><p>I’d love to wrap this up with a neat business lesson, but it wasn’t that tidy.</p><p>It was messy. Brilliant. Stressful. Muddy. Expensive. Funny. Exhausting. Occasionally terrifying.</p><p>And I loved it.</p><p>I loved designing the brand people bought into, the website that helped them take the leap, and the courses they remembered years later. I loved watching nervous runners become different people by the finish line. I loved the Reapers, the crew, the chaos, the stupid problems, the clever fixes, and the strange little community that grew around a race with a deeply unmarketable name.</p><p>The Suffering taught me that people don’t just want easy wins.</p><p>They want stories.</p><p>They want to earn something.</p><p>And for five years, we got to build that.</p><p><strong>One hill, one obstacle, one Reaper, and one slightly questionable idea at a time.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/820/0*VnXNro01dFbYd5yY" /></figure><p>Thanks for reading.</p><p>I write about the things I’m curious about and the things I’m building. Sometimes that’s service design, AI, product development and security. Other times it’s productivity, martial arts, meditation, or lessons from everyday life.</p><p>This is my little corner of the internet where I try to make sense of interesting ideas, share what I’m learning, and occasionally challenge my own assumptions along the way.</p><p>If you’d like to follow what I’m working on or explore some of the projects I’m building, you can find me here.</p><p>You can learn more at:</p><ul><li><a href="http://www.hueylabs.com/">HueyLabs</a></li><li><a href="https://drewtavernier.medium.com/">Medium</a></li><li><a href="https://www.linkedin.com/in/drew-tavernier/">LinkedIn</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5cee9a2c0f03" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[If You Can Think It, You Can Build It… Badly]]></title>
            <link>https://engineering.dunelm.com/if-you-can-think-it-you-can-build-it-badly-2bed8892db59?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/2bed8892db59</guid>
            <category><![CDATA[cursor]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[deep-work]]></category>
            <category><![CDATA[youtube]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Thu, 02 Apr 2026 07:54:58 GMT</pubDate>
            <atom:updated>2026-05-12T21:51:51.813Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>The thing they miss out is, vibe coding kinda sucks!</strong></p><p>I’ve been vibe-coding for a while now on various small projects. They work okay locally in Docker and do what I need them to do.</p><p>I find with all things, I need a project to really understand how they work. I can read about tools and frameworks, but until I’m actually trying to build something with them, it never fully clicks.</p><p>So about a year ago, I started work on a video creation tool to support my YouTube meditation channel, <a href="https://www.youtube.com/@Zen_Mountain">Zen Mountain</a>. I cobbled something together in OpenAI Codex that took the leg work out of compiling videos in Premiere, and I was happy with it for a while.</p><p>But then a new channel was in the works: <a href="https://www.youtube.com/@deepworksessions">Deep Work Sessions</a>, a channel for productivity tools like Pomodoro timers. I wanted something that was less time-consuming than <a href="https://www.youtube.com/@Zen_Mountain">Zen Mountain</a>.</p><p>When planning this channel with <a href="https://www.linkedin.com/in/mrwellsedits/">Glenn</a>, my partner in crime, we looked at how to make the videos the best they could be by hand first. We sat down and crafted perfect Pomodoro timers and other tools, looked at other channels, saw the market gap, styled our products like no one else, simple and bold, and came up with our own formula.</p><figure><img alt="Deep Work Sessions Logo" src="https://cdn-images-1.medium.com/max/1024/1*1V-wdQPVtaTnxyVT59qHkg.jpeg" /><figcaption>Deep Work Sessions Logo</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zjiNg0eBI0NyJyG-14GqdQ.png" /><figcaption>Example of thumbnails</figcaption></figure><p>The thing that struck me as I made the first few videos was that this was really a collection of elements at varying lengths stitched together.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_YW62q8Jws0m2lO-0T5U7A.png" /></figure><p>Which meant it could be systemised.</p><p>And I’d already made a simple video creation tool for Zen Mountain that incorporated FFMpeg audio &amp; video compiling so I was 20% of the way there.</p><h3>The Bit Where It Went Wrong</h3><p>At this point I had been playing with Cursor, with varying degrees of success. Smaller simple apps were easy, but as soon as they became bigger, more complex, they would spiral into a world of pain.</p><p>My v2 of the Zen Mountain tool went extremely badly. I wasted a month chasing bugs, and failed tests, but that failure gave me the foundation for the Deep Work Sessions app.</p><p>I realised two things on that journey.</p><p><strong>First, </strong>letting vibe coding tools do their thing without a solid plan means you produce a very bad product.</p><p><strong>Second, </strong>coding in languages you don’t understand means you also produce a very bad product.</p><p>That was the real shift for me.</p><p>The problem wasn’t AI. The problem was how I was using it.</p><h3>What Actually Helped</h3><p>So I went away and did some research. What helps you build with Cursor?</p><p>The answer, like with most things, is boundaries.</p><ul><li>Use Svelte and SvelteKit, languages I had learned</li><li>Use the design system I had created</li><li>Plan everything, discuss everything, and use Cursor as an assistant, not an creator</li><li>Use Cursor rules to force the behaviour you want</li></ul><p>The big one for me was moving from the tech stack that Cursor recommended, but I knew very little about, React, Next.js and so on, to one I actually understood.</p><p>I spent a month learning the basics of Svelte and SvelteKit, building on my existing knowledge of HTML, CSS and JavaScript. That meant I had far more control over what we were building, and I say “we” intentionally.</p><p>I was guiding the ship, not being taken on a journey by the wind.</p><p>It took time to get to a place where Cursor was working with me, not against me, but now I’m flying.</p><h3>What I Built</h3><p>And what did I create?</p><p>Thanks to Cursor’s Planning Agent I was able to create detailed plans, discuss issues, challenge the agents assumptions and create the Deep Work Sessions app; which is a full video creation and scheduling tool that uses AI to write the YouTube metadata and APIs to post directly to YouTube.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yt79g8DFajqHt9XbyGZzzw.jpeg" /><figcaption>Screens from the app</figcaption></figure><p>I can create a video in seconds and process it overnight.</p><p>It supports multi-project creation, has a render queue, integrates AI as part of the workflow, and uses external APIs to post to YouTube.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/989/1*Y4cY6C2dhI50Gk1jLqGmyw.jpeg" /></figure><p>I’ve deliberately kept this as a local Docker app so I can share it with Glenn and not have to worry about system compatibility issues.</p><p>It’s super MVP because it doesn’t need to be a venture capital startup, it doesn’t need to be over-engineered, it just needs to solve the problem well.</p><h3>So, What Have I Learnt?</h3><p>Vibe coding does suck.</p><p>But it only really sucks when you let it think for you.</p><p>If you let the tools run wild without a plan, you’ll get a mess. If you build in languages you don’t understand, you’ll get a mess faster. If you treat AI as the expert, you’ll usually end up somewhere you didn’t mean to go.</p><p>But if you bring structure, boundaries, taste, and at least some technical understanding to the table, it becomes genuinely useful.</p><p>That’s the bit people don’t say enough.</p><p>Vibe coding is not magic. It’s a tool.</p><p>And like any tool, it only works well when you know what you’re doing with it.</p><h3>Follow my YouTube channels here</h3><ul><li><a href="https://www.youtube.com/@deepworksessions"><strong>Deep Work Sessions</strong> </a>— Productivity tools</li><li><a href="https://www.youtube.com/@Zen_Mountain"><strong>Zen Mountain</strong></a> — Meditation channel</li><li><a href="https://www.youtube.com/@notasifu"><strong>Not A Sifu</strong></a> — My reflections on Wing Chun</li></ul><h3>Want to know more, reach out here</h3><p><a href="https://www.linkedin.com/in/drew-tavernier/">https://www.linkedin.com/in/drew-tavernier/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2bed8892db59" width="1" height="1" alt=""><hr><p><a href="https://engineering.dunelm.com/if-you-can-think-it-you-can-build-it-badly-2bed8892db59">If You Can Think It, You Can Build It… Badly</a> was originally published in <a href="https://engineering.dunelm.com/">Dunelm Technology</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building Better Communicators]]></title>
            <link>https://engineering.dunelm.com/building-better-communicators-095e521cdf88?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/095e521cdf88</guid>
            <category><![CDATA[learning-design]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[apprenticeship]]></category>
            <category><![CDATA[service-design]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Wed, 01 Apr 2026 15:34:21 GMT</pubDate>
            <atom:updated>2026-04-29T10:36:23.397Z</atom:updated>
            <content:encoded><![CDATA[<p>When I started my Service Design apprenticeship, I knew there’d be a moment where I’d need to take everything I was learning and apply it back into Dunelm. What I didn’t know was that this moment would involve me frantically designing a full workshop over a two week period, while doing my day job, discovering the limits of AI in the most philosophical way possible, and realising (again) that communication is the linchpin of… well, pretty much everything.</p><p>More specifically, it became a chance to build something genuinely useful for Dunelm. Not just because the apprenticeship said I should, but because there was a clear opportunity to create something that could help the Product Design team in a very real, very immediate way.</p><p>This is the story of how I went from <em>“Sure, I’ll create some internal training”</em> to <em>“Here’s a fully structured workshop with learning outcomes, behavioural change goals, post‑session evaluations, and somehow also a roadmap for four more courses”.</em></p><h3><strong>Where It All Began: The Brief That Wasn’t Really a Brief</strong></h3><p>The apprenticeship comes with an expectation: don’t just learn, use the learning.</p><p>So, I went off to find the area in Product Design where that learning could be most useful at Dunelm. After a few conversations with Matt, my manager, and Ryan, a fellow apprentice and Dunelm colleague, the priority became obvious: communication.</p><p><strong>Why communication?</strong></p><p>Because it sits right at the centre of how we work at Dunelm.</p><ul><li>Presenting research and ideas.</li><li>Facilitating workshops.</li><li>Influencing decisions.</li><li>Working with senior stakeholders.</li><li>Navigating the occasional tricky conversation.</li></ul><p>Even trying to explain, calmly and professionally, why the button absolutely cannot be moved there.</p><p>And when communication breaks down, good work struggles to land. But when it improves, everything else gets a bit better. Faster. Smoother. Less painful.</p><p>So rather than spreading the learning thinly across loads of topics, it made more sense to go deeper on one skill that could benefit the team straight away.</p><h3><strong>The First Attempt: “What if AI Builds It For Me?”</strong></h3><p>I tried to be clever and asked AI tools to draft some training materials. Speedy? Yes. Useful? Not… really.</p><p>What I got back were the kind of generic courses you’d expect to get in a Tuesday afternoon webinar where someone reads slides at you for an hour.</p><p>This was the moment I realised the real problem wasn’t content creation. It was learning design.</p><p>So instead of using AI to generate a workshop, I built a workflow, the Learning Designer Agent, that uses AI to challenge and improve my design decisions. Less “write this for me”, more “tell me why this doesn’t work”.</p><p>Once that was sorted, the real work began.</p><p>For those interested here’s the Learning Design Agent process, Stage 0 onwards, just to show it’s not all smoke and mirrors.</p><figure><img alt="The Learning Designer Process" src="https://cdn-images-1.medium.com/max/1024/1*NntLUqizg42jT-mYF6xjTA.png" /></figure><h3><strong>Designing the Workshop: A Crash Course in Structure, Intent, and Mild Panic</strong></h3><p>I followed an iterative approach inspired by the Double Diamond. Discover, Define, Develop, Deliver, only with fewer diamonds and more caffeine.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*upoKCbj7zqRFcPphH6VYNw.png" /></figure><h4><strong>Discover</strong></h4><p><strong>What exactly is going wrong with communication?<br></strong> Spoiler: quite a lot. Body language, tone, pacing, and overly complicated wording were all culprits.</p><h4><strong>Define</strong></h4><p>The workshop couldn’t just be “here are some tips”. It needed behavioural outcomes. People should do something differently afterwards, not just nod and forget everything by lunchtime.</p><h4><strong>Develop</strong></h4><p>This is where the fun started.</p><p>I created demonstrations, partner exercises, guided reflections, and questions designed to surface new awareness. Everything had to be practical, fast, and safe… no role play nightmares.</p><p>The whole thing was built around three core “signals” of communication:</p><ul><li>Body language (what people see)</li><li>Tone of voice (what people hear)</li><li>Words (what you say)</li></ul><h4><strong>Deliver</strong></h4><p>Two weeks later (somehow), Communication 101 was born.</p><h3><strong>Running the Session: Or… How I Discovered People Love a Bit of Chaos</strong></h3><p>The live workshop was energetic, hands-on, occasionally surprising, and, thankfully, well received.</p><p>Participants learned:</p><ul><li>How mismatched body language completely derails a message</li><li>Why tone gives you away before your words do</li><li>How pacing affects trust</li><li>That saying less is often clearer and more confident</li><li>And that simple, direct words aren’t “dumbing down” they’re “helping people understand you”</li></ul><p>They also left with one behaviour to try in a real conversation that week. Not a sweeping transformation. One specific thing and that’s how we reinforce change.</p><h3><strong>What the Feedback Said (and Why It Meant a Lot)</strong></h3><p>Feedback was collected at two points:</p><ul><li>24 hours after the workshop</li><li>Two months later (not collected yet)</li></ul><p>From the 24 hour feedback, a few themes came through clearly:</p><p><strong>People became more aware of what really shapes a message.<br></strong>Body language and tone suddenly made sense in a new way.</p><p><strong>People identified clear behaviours to improve<br></strong>Eye contact. Slowing down. Speaking less but saying more. Matching tone to intent.</p><p><strong>The interactive design worked.<br></strong>The demos, paired activities, and questions all landed well.<br>Some wanted more time to thoroughly explore the ideas though.</p><h3><strong>What I Learned Along the Way</strong></h3><p>This project quietly re‑wired how I think about training design. A few things stood out:</p><ul><li>Training need’s structure, not just content</li><li>Behavioural change is the real measure of success</li><li>SME support makes everything sharper</li><li>AI is great as a critique partner but terrible at writing human training on its own</li><li>And finally: You can deliver great training in a short amount of time if you have structure, intent, and slightly unhealthy determination</li></ul><h3><strong>Where This Goes Next</strong></h3><p>What started as one workshop has now become the start of a wider internal communication series for Dunelm:</p><ul><li>Communication 101</li><li>Building Stronger Relationships</li><li>Presenting With Impact</li><li>How to Effectively Challenge</li><li>Written &amp; Digital Communication</li></ul><p>So, this ended up being more than just a one-off apprenticeship requirement.</p><p>It became a way to create something useful for Dunelm now, while also building a foundation the team can keep using and developing over time.</p><h3><strong>Final Thoughts</strong></h3><p>This project began as a requirement of my apprenticeship. It ended up shifting how I approach training design, communication, and learning altogether.</p><p>But more than that, it became a chance to build something genuinely useful for Dunelm. Something practical. Something people could apply immediately. Something that helps good work land better.</p><p>It wasn’t about making a workshop quickly. It was about making a workshop that worked.</p><p>And based on the feedback, the behaviour changes, and the follow‑up results… it did.</p><p>If anything, it proved the point I started with:<br><strong>When you improve communication, you improve everything else.</strong></p><h3>Want to know more, reach out here</h3><p><a href="https://www.linkedin.com/in/drew-tavernier/">https://www.linkedin.com/in/drew-tavernier/</a></p><h3>Follow my YouTube channels here</h3><ul><li><a href="https://www.youtube.com/@deepworksessions"><strong>Deep Work Sessions</strong> </a>— Productivity tools</li><li><a href="https://www.youtube.com/@Zen_Mountain"><strong>Zen Mountain</strong></a> — Meditation channel</li><li><a href="https://www.youtube.com/@notasifu"><strong>Not A Sifu</strong></a> — My reflections on Wing Chun</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=095e521cdf88" width="1" height="1" alt=""><hr><p><a href="https://engineering.dunelm.com/building-better-communicators-095e521cdf88">Building Better Communicators</a> was originally published in <a href="https://engineering.dunelm.com/">Dunelm Technology</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Portable User Testing Studio]]></title>
            <link>https://drewtavernier.medium.com/portable-user-testing-studio-929bf5ae00e9?source=rss-1714908ae96a------2</link>
            <guid isPermaLink="false">https://medium.com/p/929bf5ae00e9</guid>
            <category><![CDATA[notetaking]]></category>
            <category><![CDATA[testing]]></category>
            <category><![CDATA[user-research]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[portable]]></category>
            <dc:creator><![CDATA[Drew Tavernier]]></dc:creator>
            <pubDate>Wed, 19 May 2021 11:30:48 GMT</pubDate>
            <atom:updated>2021-05-19T11:36:54.680Z</atom:updated>
            <content:encoded><![CDATA[<p>A Cost Effective Approach</p><p><strong>At Dunelm, we have been working on setting up a portable studio for a while. We spent months researching various solutions, only to hit dead ends on cost, set-up time or portability.</strong></p><h3>Why do we need a studio?</h3><p>The primary factor was cost saving. Running in-depth testing using a viewing studio, mediator and receiving reports a week after testing mounts up. A two-day test costs the business upwards of £10,000. With this outlay per session, it makes companies reluctant to test regularly. We have reduced this to below £1,000. There is a breakdown of costs at the end of this article.</p><p>Money couldn’t be the only driver for having the studio, so before we started we set ourselves a definition of success, these include:</p><p><strong>Time</strong></p><ul><li>Getting to the data quicker</li><li>Being able to collate our conclusions in quickly and easily</li></ul><p><strong>Set-Up</strong></p><ul><li>One man set-up</li><li>Fits in a car boot</li><li>Connects portable devices</li><li>Has multiple viewing angles</li><li>Can be done in any city</li></ul><p><strong>Team</strong></p><ul><li>Allows observation from another location</li><li>Actively encourages contribution and recommendations from the team</li></ul><h3>It’s Not As Easy As Hiring Someone In</h3><p>Running your testing saves money, there is no doubt about that, but it also adds additional workload on your team. The bulk of the costs from our external testing companies were the Recommendations Reports that were produced in the weeks after the tests. Now we rely on the project team to form hypostasis within the testing.</p><p>To help with this, we run active listening note taking during the sessions to keep chatter and distractions to a minimum. You can <a href="http://drewtavernier.com/articles/active-listening-note-taking.php">learn more about our observations techniques in this article.</a></p><h3>The Testing Studio</h3><p>We carefully pieced this solution together from various sources. One of the key stars of the show is the Startech KVM Console to Portable Laptop Adaptor. Initially designed to access rack servers without the need for a monitor per rack. This bit of kit is excellent and allows us to connect Apple products to the system which enables us to view the device on the laptop screen. We’re currently working on options for Android.</p><p>The laptop records all the data; sound, screen, and customers face. When an external device is connected, it showed on the laptop’s screen and captured with the rest of the data.</p><p>Multiple angles are captured using two Logitech, C920 HD Pro Webcams and shown in the viewing room. In our current iteration, we don’t record this data but in the future, we may.</p><p>We use the multiple angles in the viewing room to observe not only the customer but the interviewer. The extra angels prove invaluable as we can view the body language from both parties.</p><figure><img alt="Moderator and participant using the portable studio" src="https://cdn-images-1.medium.com/max/720/0*McWRvR083bKo9_x9.jpg" /><figcaption><em>Moderator and participant using the portable studio</em></figcaption></figure><h3>The Viewing Studio</h3><p>Instead of a viewing studio, we use hotel conference rooms. The key here is having two that are next to each other.</p><p>Our viewing screen is a 42 inch 4k TV. The screen is big and detailed enough that everyone in the room can view it and the sound is clear enough that the volume isn’t excessive.</p><p>A second computer monitor has the two additional cameras we use to view body language and social interactions. We found this especially useful in the initial interview stage where we find out the customer’s background, usage habits and relationship with the technology.</p><figure><img alt="Active listening note taking in our viewing room" src="https://cdn-images-1.medium.com/max/720/0*SZai3IiBP3LsJfk3.jpg" /><figcaption><em>Active listening note taking in our viewing room</em></figcaption></figure><figure><img alt="Notes from our active listening exercise" src="https://cdn-images-1.medium.com/max/720/0*hZKfNRFUzm4oiURM.jpg" /><figcaption><em>Notes from our active listening exercise</em></figcaption></figure><h3>Does It Work?</h3><p>The short answer is yes.</p><p>And now for the long answer; running our testing helps us look at the bigger picture. We understand that an interview stage is an under-utilised tool.</p><p>We ground our hypothesis in the personality type in front of us, this adapts and expands as multiple people come through giving us a broader picture.</p><p>As a group, we pick up on trends as they happen and give ourselves time to discuss, analyse and dissect directly after each session.</p><p>Over the day of the test, we build a detailed view of the failures and successes. There is no need to wait for a week for a report because the report writes itself with buy-in from all those taking part.</p><p>Do we believe we made the right choice?</p><p>Hell yeah, and here are some figures to help you get you heading on the highway to awesomeness</p><h3>The Figures</h3><p>Our traditional route was going through a testing company and hiring a viewing studio at a total cost of £9,000. Here is a breakdown of costs:</p><ul><li>£1000 per day to hire of viewing studio</li><li>£1000 per day for an external mediator / interviewer</li><li>£1000 for recruitment</li><li>£5,500 for a report generation</li><li>£500 enumeration</li></ul><p><strong>Our portable studio cost</strong></p><ul><li>£4,953.60 as a one-time kit purchase.</li></ul><p><strong>Cost per test (£840)</strong></p><ul><li>£350 for hiring 2 conference rooms</li><li>£290 recruitment cost</li><li>£200 incentive cash, plus any expenses</li></ul><p><strong>Saving £3,706.40 per test for the first test and £8,660 per test there after.</strong></p><p><em>Originally published at </em><a href="http://drewtavernier.com/articles/portable-testing-studio.php"><em>http://drewtavernier.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=929bf5ae00e9" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>