Agentic engineering, made practical.
Agentic engineering consultancy
Your developers already have Copilot or Claude Code, yet delivery hasn't changed. Codomain measures how ready your projects are for coding agents, fixes what holds them back, and runs a software factory on your real backlog.
Led by Jakša Vučković, PhD in distributed systems, 20 years of software consulting.
Why bother
Agents don't just type faster. They change the economics of software, and with them the decisions a technology strategy is built on.
Agents work around the clock, and you can run as many in parallel as your codebase and your reviewers can absorb. On well-prepared codebases we've seen order-of-magnitude gains in throughput, and far more on well-bounded work.
When rebuilding is cheap, architecture stops being a decade-long commitment. Trying an alternative design, or migrating to a new stack, can be an overnight experiment rather than a multi-year programme.
Tests, documentation, and keeping them up to date: agents do them every time, because it's part of the task. Your codebase gets better as a side effect, for people as well as for agents.
The upgrades, refactors and missing tests that never won a place on the roadmap can now run in the background, in parallel with feature work.
The approach
The same agent that ships a story overnight in one codebase creates a week of rework in another. The difference is rarely the model. It's whether the agent can find the context, reproduce the environment, and check its own work.
Reproducible dev environment, tests the agent can run, requirements and architecture it can read. Agents plan, implement and verify whole stories while people review outcomes.
No feedback loop, knowledge that lives in people's heads, a setup nobody can automate. Agents guess, regressions slip through, and the team concludes "AI doesn't work here".
Our two maturity models — Readiness (is the codebase ready?) and Adoption (how does the team use agents?) — scored against your repos with automated checks, so progress is a number, not an opinion.
No systematic agent use.
Agents assist individual developers, step by step.
Agents implement whole stories; people review the diff.
Agents continuously pull work from the backlog. People supervise.
Tech debt, security and dependencies managed by agents too.
From the Agentic Adoption Maturity Model. Progress is re-measured throughout, not assumed.
Services
We work on time and materials, with a rough estimate up front. Most clients start with the assessment and build from there.
We score every repository against the readiness criteria and your team against the adoption model, then tell you what is holding agents back.
We close the gaps the assessment found. Most of the work is done by agents under our supervision, which also shows your team how it's done.
We install Microfactory on your team's real Jira or GitHub backlog and run it with them until it runs without us.
Rolling this out across many teams? We act as your fractional head of agentic engineering: strategy, rollout plan, team-by-team adoption and the metrics to show it's working. Ongoing, with quarterly adoption targets.
How we work
We re-run the same automated assessment throughout the engagement. You see the progress, not a slide deck about it.
We bill time and materials. You get a rough estimate up front and see the time spent every week. Priorities shift as we learn what pays back, and you decide what you want us to fix, what you want to fix yourself or stop at any time.
Our tooling runs inside your repositories with your own Claude Code or Copilot subscription and your existing Jira or GitHub. No platform to buy, nothing to host.
Every contract includes a licence to use and adapt our frameworks and tooling inside your organisation, and it stays valid after we leave. We aim to make ourselves unnecessary.
Our toolkit
We developed this toolkit for our own work and use it daily on our own projects. It isn't sold as a product. It comes with us, gets tuned to your codebase during the engagement, and is licensed to you with every contract.
Conceptual frameworks for modelling agentic maturity: how ready a codebase is for agents, and how far a team has adopted them. The common language behind every assessment and roadmap.
Tools that analyse how agent-ready a project is, and re-measure it as the work progresses, so improvement is visible rather than anecdotal.
Agents that improve your codebase for agents. The same changes also make life better for human developers and lift overall quality.
A software development lifecycle for agentic development, proven on real projects, with humans supervising where it matters. Microfactory is the platform that runs it.
About
Codomain is run by Jakša Vučković, a computer scientist with a PhD in distributed systems from the University of Bologna and 20 years of software consulting. He spent over fifteen years at Zühlke Engineering delivering and leading projects for banks, rail, energy and research organisations, then served as Head of Engineering at Alvie and Director of Engineering at the Royal Pharmaceutical Society.
Most recently he shaped the AI-assisted development strategy for a UK retail technology organisation and coached its engineers and managers through adoption. He wrote the readiness and adoption maturity models and builds the toolkit Codomain brings to every engagement.
More on LinkedIn.
FAQ
Time and materials. We give you a rough estimate up front and you see the time spent every week. You can scale the engagement up or down, or stop, at any time.
Claude Code and GitHub Copilot. The maturity models are agent-agnostic; our tooling runs on Claude Code.
No more than it already does with your agent subscription. The tools run inside your repositories with your accounts. There's no Codomain platform, and no credentials are stored in configuration.
That's exactly what the assessment is for. Putting a factory on a codebase without tests or a reproducible environment makes things worse, so we tell you if that's the case and help you fix it.
Jira and GitHub Issues. For small projects, a plain TODO.md in the repo works too.
You own your code and everything we change in it. Our frameworks and tooling stay ours, and every contract includes a licence to use and adapt them inside your organisation.
Yes, with clients across Europe. On-site sessions for kick-offs and workshops can be arranged.
Contact
Tell us about your teams and codebase. We'll tell you honestly how we can help.