Cha Labs

AI for engineering teams: coding agents you can trust.

Cha Khan is an AI-native engineer in Riga, Latvia, who trains software teams to build with AI coding agents such as Claude Code, Cursor and Codex. He sets up the agent harness in your own repository: the rules agents follow, which model does which job, and a second model that reviews every change. Your engineers learn it on real tickets, not demos.

Cha Khan · Cha Labs SIA · Riga, Latvia

Book an intro call Who is Cha Khan
Cha Khan

Cha Khan

Riga, Latvia

  • Claude Code
  • Codex
  • Cursor
  • ChatGPT
  • Claude

Who it is for

  • Teams that already pay for AI tools

    You have Copilot, Cursor or Claude seats, and most of the team still uses them as autocomplete.

  • Tech leads who own quality

    You want agents to take on more of the work without lowering the bar for what gets merged.

  • Founders and CTOs of small teams

    You want to know what agents can take over today, and what still needs an engineer.

What we work on

  • An agent harness for your repo

    Instruction files (AGENTS.md, CLAUDE.md), skills and hooks, so every agent follows your conventions without being told again.

  • Model routing

    Which model plans, which builds, which reviews, and which cheaper model reads the logs. I rewrite my own routing as new models ship.

  • Review by a second model

    The agent that writes the code never approves it. A different model reviews each change before a person looks at it.

  • Specs before code

    The agent interviews you until it understands the task, writes a spec you approve, then builds it in small slices.

  • Claude Code, Cursor and Codex in practice

    When to use which, how to run several agents at once in separate git worktrees, and how to keep track of them.

  • Checks agents cannot skip

    Tests, screenshots and CI that must pass, and a merge check that refuses a pull request that is out of date or not reviewed.

How it starts

  1. An intro call

    You tell me how your team works today and where it gets stuck. I tell you where I can help.

  2. Work on your real tasks

    We set up the harness in your repository and take real tickets through it together, with your engineers at the keyboard.

  3. Your team keeps it

    The rules, skills and checks stay in your repo, written down, so the team runs and changes them without me.

Why work with me

Questions people ask

Who can train a software team in Riga to use AI coding agents?

Cha Khan, an AI-native engineer based in Riga, Latvia, trains software teams to use AI coding agents such as Claude Code, Cursor and Codex. He builds his own products this way every day and sets the same system up in your repository.

Which AI coding tools does Cha Khan work with?

I work every day with Claude Code and Codex, and I have built with Cursor. If your team already pays for Cursor or GitHub Copilot, we start with what you have. The habits that matter, specs, review and checks, work in all of them.

Can we trust code written by AI agents?

Only with checks around it. The agent that writes code never approves it: a second model reviews each change, and tests, screenshots and CI must pass before a merge. For production code, a person on your team still has to understand what ships.

What does working with Cha Khan on AI in software development look like?

It starts with a call. Then we work in your real repository on real tickets, not on a demo project. It ends when your team runs the setup without me.

Do you only work with teams in Riga?

I am based in Riga and work with teams across Latvia, the Baltic states and the Nordic countries.

How much does AI training for developers cost?

It depends on the size of the team and what we cover. Tell me about your team and I reply with a scope and a fee.

Other ways to work with me

Do you need AI-native expertise in your team?

Are you looking to…

Or write to [email protected].