The agency modelis finished.
We are not a leaner version of what came before. We are what replaces it: one senior team, fluent in models and infrastructure, shipping what used to take thirty people.
Five beliefs. Everything we build comes out of them.
Small is the advantage, not the compromise.
Every layer between the thinking and the building is a tax you pay in weeks. No account managers. No handover decks. No approval chain. You talk to the people writing the code, and the distance between a decision and a shipped change is measured in hours.
Speed is a quality signal.
Slow work is rarely careful work, it is usually work waiting on someone else. Six days to a working pilot. Three weeks to production. The point isn't the clock: it's that you learn whether the idea is real before the budget decides for you.
The machine does the work. The judgement is ours.
AI handles research, scaffolding, documentation, tests, the part that used to eat the first month. What it cannot do is decide what deserves to exist, or recognise the architecture that will quietly fail in year two. That is twenty years of pattern recognition, and it is the part you are actually hiring.
We will argue with your brief.
A studio that builds exactly what it is told is a very expensive pair of hands. If the idea won't survive contact with the market, you should hear it from us in week one, not from your users in month nine. Occasionally the honest answer is: don't build this yet.
You own all of it. Including the parts that make us replaceable.
Code, design, infrastructure, the agent prompts, the deployment pipeline. Fixed fees, dates agreed before we start, a weekly note on where things stand. No retainer that quietly becomes permanent, no licence you can't walk away from.
One senior team, and the machines that make it look like thirty.
No account layer, no bench to keep busy, no work passed down a chain until nobody remembers why it started. The people who scoped it are the people who build it.
The same brief, two decades apart.
Averages across twelve recent engagements. Nothing was cut. The month that used to go on paperwork now goes on the product.
Paul Gosnell
Twenty years of shipping for startups, enterprises and the occasional government. Grew an agency from one to thirty, took it through a pandemic, and came out with a conviction that has only got stronger since.
Smaller teams, better tools, faster results. p0stman runs as an AI-native studio with a curated network of specialists, designers, engineers, strategists, pulled in when the work needs them. No bloat. No bench. Experienced builders with a great deal of leverage.
Paul Gosnell · Founder
AI agents & intelligent systems
Voice and chat agents in production, not demos. Multi-model orchestration. Workflow automation that removes forty hours a month.
02MVPs & digital products
Booking platforms, SaaS dashboards, mobile apps, commerce. Six-day pilots to find out if it's real, three-week builds to launch it.
03Fractional CPO & CTO
Product leadership on demand. Transformation roadmaps, team scaling, white-label builds for agencies who need the bench.
04Agent-ready web
MCP servers, A2A endpoints, structured context. Making sure the next generation of buyers, which is software, can find and use you.
What's the ambitious version?
Most founders think small because budgets and timelines force them to. That constraint is gone. Tell us what you'd build with a hundred experts and six months, then let's do it in three weeks.