Fixed-scopeAI sprints.
In 4 to 8 weeks I build 2 to 3 production AI agents or automations for one specific bottleneck in your business. Documented, handed over, running. No hourly billing, no scope creep.
A good fit
B2B SaaS founders and teams with one clear, repetitive bottleneck.
Non-technical founders who want to ship with AI without hiring a team.
Teams with no spare engineering capacity for internal tooling.
People who care about outcomes running in production, not a demo.
Not a fit
You want a chatbot demo or a proof of concept for a pitch.
You need a full custom SaaS product built from scratch.
You're looking for ongoing staff augmentation by the hour.
The problem isn't defined yet and there's nothing to automate.
- Week 0Scoping call
We pick one bottleneck worth automating and agree the exact scope and a fixed price before anything starts. No surprises later.
- Weeks 1–2First agent
I build the first production agent or automation and wire it into your existing stack. You see it running early, not at the end.
- Weeks 3–6Build out
I build the remaining agents and connect the flow end to end, so there are no manual steps left between stages.
- Final weekHandover
You get the full source code, documentation and a walkthrough. Everything runs on your infrastructure and you own all of it.
2 to 3 production systems
Agents or automations running on your infrastructure, not a prototype.
The full source code
You own everything. No lock-in, no black box, nothing hidden.
Documentation & handover
A written walkthrough and a live handover so your team can run and extend it.
A modern, proven stack
Built with the Vercel AI SDK, Claude and the tools your stack already uses.
Fixed scope, fixed price, agreed before we start.
Every sprint is scoped to one bottleneck and priced up front, so you know exactly what you get and what it costs before any work begins. The figure depends on how many agents we build and how many systems each one plugs into.
No hourly billing, no creeping estimates. We settle the price on the scoping call once the problem is clear. Right now I take a few first clients at founding pricing, in exchange for turning the results into a case study.
You start with almost no risk.
The intro scoping call is free, so we sharpen the scope together before you commit. Then a paid process-mapping step finds what's actually worth automating for you. If it doesn't show a clear saving, we stop there and you pay nothing further. If we go ahead, the mapping fee counts toward the sprint.
I run this solo, and I plan for it.
One person means continuity is a fair question, so here's how I keep it from being a risk. I hold my client load deliberately small, so you get real availability instead of a place in a queue. Everything I build is documented and handed over to run on your own infrastructure, so nothing important lives only in my head. And if I'm ever overloaded or away, your systems keep running without me, because they're yours and documented well enough for another developer to pick up. You're buying a working system you own, not a dependency on me.
01How long does a sprint take?
Between 4 and 8 weeks, depending on how many agents we build and how complex the integrations are. We fix the timeline on the scoping call.
02What if I don't have a technical team?
That's a common case. I build and ship without needing a separate dev team, and I hand over documentation and a walkthrough so a non-technical founder can keep it running.
03Who owns the code?
You do. You get the full source code and it runs on your own infrastructure. There is no lock-in and nothing hidden.
04What happens if an agent fails in production?
Agents are built with monitoring and reporting, so failures are visible rather than silent. After handover you can run and fix them yourself, and I document how.
05What does a sprint cost?
Each sprint is scoped and priced up front against one bottleneck, with the figure driven mostly by the number of agents and how complex the integrations are. We agree the fixed price on the scoping call. For the first few clients I'm running founding pricing in exchange for a case study.
06What stack do you build on?
The Vercel AI SDK, Claude and the tools your stack already uses. The goal is a system that fits your infrastructure, not a new silo to maintain.
07How is my data kept secure?
Everything runs on your own infrastructure and under your own keys, so your data stays in your systems, not mine. I only need access during the build, and you revoke it at handover.
08What if the sprint doesn't deliver a working result?
You don't commit to a build blind. The intro scoping call is free, and a paid process-mapping step first confirms what's worth automating. If it doesn't show a clear saving, we stop there and you pay nothing further. If we go ahead, the mapping fee counts toward the sprint, and the goal is a result running in production, not a report.
09What do you need from me to start?
The scoping call, access to the one system the bottleneck touches, and someone who can answer questions along the way. That's it. The building is on me.