I take software teams from "we tried Copilot" to an agentic workflow that runs on their own stack, with their own people and their legacy code. I've done it at a 52-person software company and at a 12-developer backend startup.
Two weeks with a fixed scope, spent on your codebase, your team and your bottlenecks.
A few sprints on a real slice of your product, measured by outcomes rather than hours.
Roll it out across teams and keep it working after I step back.
I picked agentic workflows for the gamedev and full-stack teams, built and tested custom skills for each stack, and trained the leads to run them.
Agents now write the code and manage the tickets. Everyone uses one workflow with quality controls, and the team had workshops to learn it.
To see how I build agent tooling, look at claude-plugins, my open-source Claude Code plugins. More client work is on the work page.
Your team stops using AI as a smarter autocomplete and starts running an agentic workflow: agents write the code, and your leads own the process around them. I set up the workflow, the custom skills and the quality controls, and I train the people who will run it.
Yes. I do this work on real legacy codebases as well as greenfield projects. I build and test the custom skills against your stack and conventions, so the agents work with the code you actually have.
With a two-week discovery, fixed in scope. I look at your codebase, your team and your bottlenecks, and you get a staged plan out of it, whether or not we continue.
Documented workflows your team owns, custom skills for your stack, and leads who know how to run an AI-first team without me.
At a 52-person international software company, across gamedev and full-stack teams, and at a 12-developer backend startup where agents now write the code and manage the tickets.
Tell me what you're working on and what "better" would look like. I'll reply with whether I can help and how we'd start. The price depends on the scope, so we agree on it once we know what the work is.