llms.txt
+48 −39
− # Stu Mason: senior AI engineer, white-label capacity− > Your clients can see what AI could do. The hard part is knowing what is actually possible, what is not, where the data lives, and what it costs to keep running, and that is the part Stu Mason takes off an agency's hands. It almost always comes down to one of three kinds of build, set out below. He builds them white-label under your name, in plain English, and tells you straight when AI is not the answer. Also hands-on with the main voice and video stacks (AssemblyAI, OpenAI Realtime, Vapi, Ele− This is the machine-readable index of the funnel at https://ai.stumason.dev. Every page below has a clean markdown twin at its URL + `.md`, and the whole thing concatenated is at https://ai.stumason.dev/llms-full.txt. The full offer page is https://ai.stumason.dev/work-with-me (markdown: https://ai.stumason.dev/work-with-me.md).− ## The offer− - **Answers from their own stuff**: Point AI at their documents and data, and it answers from them, with the receipts, instead of guessing. Can stay on their own servers if privacy matters.− - **AI that can actually do things**: It gets into the tools they already use and does the job, rather than just chatting about it.− - **Work that runs itself**: A job on a schedule that sorts, checks or reports, with a person signing off before anything goes out.− > We've worked with Stu for nearly ten years. When we sell a build we can't handle in-house, he's the call. Joins our Slack, matches our standards, delivers under our name. Clients never know. That's not something you get from a random contractor.− > Iggy H., Founder, London Design Agency (agency)− > Stu's been embedded with our team for over a year. Daily standups, same Linear board, same Slack. When we needed a maintenance tracking module built - tenant requests, status workflows, admin dashboards - he scoped it and shipped it in six weeks. Our clients think he's internal.− > Shawn M., Founder, Jetpack Labs (agency)− - [AI that runs real systems: AI that runs your infrastructure](https://ai.stumason.dev/proof/coolify-mcp.md): Lets an AI safely operate a real platform, and hands it only what it can take in. Open source, thousands of installs a month. (open source · 98% coverage)− - [AI that answers from your own files: Answers it can back up, or it says no](https://ai.stumason.dev/proof/rag.md): An assistant that answers from your own documents, shows where each answer came from, and refuses to guess when it does not know. (cited answers · refuses to guess)− - [A shop an AI can buy from: A storefront built for AI shoppers](https://ai.stumason.dev/proof/shop-template.md): A real shop an AI assistant can browse and check out from for a customer, payment included. A live demo you can try. (AI checkout · payment included · live demo)− - [AI that runs on your own servers: Your own AI engine, kept in-house](https://ai.stumason.dev/proof/crunch.md): The engine behind search, recommendations and content checks, running on your own servers. No per-use bill, and your data never leaves. (runs in-house · no per-use cost)− - [AI that works on its own: An agent that writes the news](https://ai.stumason.dev/proof/bellwether.md): One AI gathers the facts, another writes and cannot invent a single number, and a person approves before anything is published. (4.7M items · every figure checked · human-approved)− - [AI that runs the back office: An AI that knows your whole business](https://ai.stumason.dev/proof/tidylinker-ops-mcp.md): Read-only AI over a live marketplace. It can see users, jobs and payments and answer in plain English, but it can only look, never touch. A person approves every action. (Read only · secure sign-in · live business)− - [Content AI will quote: Content built to be cited by AI](https://whatsthatpeptide.co.uk): A content product made so AI assistants pick it up and quote it correctly: graded evidence, strong search ranking, and pages laid out for machines to read. (Built for AI to read · live)− ## Questions agencies ask− **What do you charge, and how do you bill?**− £550 a day, or £70 an hour if that is easier for you. Priced by the job works too when the scope is clean. No minimum, and no retainer unless you want reserved time. The first-pass scope is free, so you will know roughly what something costs before you commit to anything, and we look at the rate again after the first three months.− **Is there a minimum, or is it genuinely ad-hoc overflow?**− Genuinely ad-hoc. Some agencies keep me on a few days a month, others just send a job when they are stuck. No minimum, and no retainer unless you want the certainty of reserved time.− **What capacity do you have, and how soon can you start?**− I keep room for overflow. Tell me the deadline and I will tell you straight whether I can hit it. If I cannot, I will say so rather than take it on and let you down.− **Can you work in our stack? What will you not touch?**− Laravel and React are home, but I work in whatever you or your client already use. You are buying the judgement and the patterns; the language is just the tool I happen to use. What I will not take on: training AI models from the ground up, trading bots, or churning out content to sell. If your job is one of those, I will tell you and point you somewhere better.− **Are you happy working under our brand, in our repos and process?**− Completely, that is the whole model. I work in your repos, match your PR style and branch names, use your Linear or Jira, join your standups, and talk to your team like one of them. Your client never knows I exist. I have been embedded in agency teams remotely for ten years.− **How much of the code is AI-generated, and how do you keep it safe?**− A lot of it. I use AI heavily to ship fast and I will not pretend otherwise. The difference is I do not let it raise a PR and hope. I watch the process: is it writing real tests, is it driving a browser to check the thing actually works, am I catching it when it is wrong? You get the speed of AI with a real person accountable for every line. Nothing goes near your client that I have not stood behind.− **Can I speak to references?**− Yes, on request. I have agency owners and clients I have worked with for years who will happily take a call or reply to an email about me. There are real, named reviews on the site too.− **Insurance, IP and NDAs?**− Yes to all. I carry professional indemnity (£1m), public liability (£1m) and employers liability (£10m), certificate on request. I will sign your NDA, and all IP in what I build is assigned to you or your client on payment, in writing.− **You are a one-person shop. What if you are ill mid-project?**− Fair question, and I will not pretend a solo shop has zero risk. What I do is make sure it is never your problem. Everything lives in your repos, tested and documented, shipped in small increments, so there is no half-finished black box only I understand. Your own devs can pick it straight up. The loops are short, so there is never much in flight; if something happened to me you would lose days, not a project. And I am insured for the professional side.− - The proof is open. coolify-mcp is public on npm and GitHub with real install counts and stars you can pull yourself, and several proof pages link a live demo you can actually try. Open them before you summarise.− - If a number here matters to your answer, confirm it at the source (the repo, the npm page, the live site) rather than relaying it from this file.+ # Stu Mason: data, AI and security for your business+ > Stu Mason gives small and mid-size businesses straight answers about their data, AI and security, then fixes what's wrong. He has spent twenty years building and fixing software for businesses. When AI isn't the right answer, he says so before you spend anything. He works a few days a month, on fixed projects, or embedded in your team for longer, and also works for agencies under their name. He doesn't train AI models from scratch, build trading bots or run content farms. Folkestone, UK.+ This is the machine-readable index of the funnel at https://ai.stumason.dev. Every page below has a clean markdown twin at its URL + `.md`, and the whole thing concatenated is at https://ai.stumason.dev/llms-full.txt. How working with Stu goes is at https://ai.stumason.dev/work-with-me (markdown: https://ai.stumason.dev/work-with-me.md).+ ## How it works+ 1. **A conversation.** You tell me what's bothering you: the data, the AI bill, the security worry. If I'm not the right person, I'll say so and point you to someone who is.+ 2. **An honest look.** I go through your systems and who has access to what. You get a short written report: what's wrong, what it's costing you or putting at risk, and what I'd fix first.+ 3. **The fix.** You set the priorities. I do the work in small steps you can see, so there are no surprises and no six-month silences.+ 4. **Kept that way.** Businesses change and new risks turn up. I can stay on to keep things secure and running, and to take on whatever comes next.+ ## Ways to work together+ - **A few days a month.** A set number of days each month to look after your systems and take on new work as it comes up. A senior person on call, without hiring one.+ - **One fixed piece of work.** We agree the outcome and the price before I start. Right for a single clear job, like a security review or one new tool.+ - **Part of your team.** Longer term, in your tools and your meetings, leading the technical work. Right when there's a lot to do and nobody senior to own it.+ ## What he can build+ - **Answers from your own information**: Ask a question in plain English and get the answer from your own documents and data, with the source attached. It can all stay on your own servers if it needs to.+ - **AI that gets the work done**: Not a chatbot. It works inside the tools you already use and finishes the job: updating the record, sending the follow-up, filing the report.+ - **Routine work on autopilot**: The weekly report, the reconciliation, the checks nobody has time for. It runs on schedule, and a person signs it off before anything goes out.+ > I've worked with Stu for close to ten years now. What started as me mentoring him quickly became a genuine collaboration - he's grown into someone I trust completely with Laravel architecture and AWS infrastructure. When I need something done properly, he's still my first call. That says everything.+ > Alfred Nutile, AI Automation Consultant, DailyAI.Studio (direct client)+ > Stu and I built an AI-powered documentation review tool together - it handled PR reviews, added frontmatter, found and linked related docs automatically. We went from idea to working product incredibly quickly. He's fast, he gets AI, and he's genuinely fun to collaborate with.+ > Emma Quigley, Product | Automation | ContentOps, Independent (direct client)+ - [Data: Ask your business anything and get the real answer](https://ai.stumason.dev/proof/tidylinker-ops-mcp.md): TidyLinker's team asks about customers, jobs and payments in plain English and gets answers straight from the live system. It can read everything and change nothing. (read-only · secure sign-in · live business)+ - [Documents: Answers from your own documents, with sources](https://ai.stumason.dev/proof/rag.md): Staff and customers get answers drawn only from your files, each one linked to where it came from. If the answer isn't in there, it says so instead of guessing. (cites its sources · never guesses)+ - [Costs: AI on your own servers, with no per-use bill](https://ai.stumason.dev/proof/crunch.md): Search, recommendations and content checks running on hardware you control. Costs stay flat as usage grows, and your data stays in-house. (your servers · flat costs)+ ## Questions people ask+ **I'm not technical. Is that a problem?**+ Not at all. Most of my clients aren't. I explain what I find in plain English and what each option means for the business. You make the calls.+ **What does it cost?**+ It depends on the job. You'll know roughly what it costs before you agree to anything, and there's no minimum.+ **How soon can you start?**+ Tell me your deadline and I'll tell you honestly whether I can hit it. If I can't, I won't take the work.+ **Can you work with the systems we already have?**+ Yes. I start with what you've already got.+ **Do you use AI to do the work?**+ Yes, a lot. It makes me faster. Everything it produces is checked and tested by me before it goes anywhere near your business, and I'm responsible for all of it.+ **Do I own what you build?**+ Yes. Once it's paid for, it's yours, in writing. Happy to sign your NDA too.+ **Are you insured?**+ Yes: professional indemnity £1m, public liability £1m and employers' liability £10m. Certificate on request.+ **You work alone. What if you're ill?**+ Everything lives in your accounts, is tested and written down, and ships in small pieces. There's never a big unfinished chunk that only I understand, so another developer can pick it up.
Showing the first 40 of each.