AI-assisted development as your team's way of working

We teach software teams to work with coding agents in a way that lasts: shared instructions, reusable skills for the agents, orchestration of several agents and review the team can rely on. The aim is a practice that stays with your team after the engagement ends, not a one-off effort on a single project.

Request a proposal

Fits when

You have a development team already using AI tools, each person in their own way, or you are about to start working with coding agents in a new environment. You want that to become a shared, reviewed practice.

What is included

A review of how the team works today: codebase, pipeline, code review and where agents are already in use.
Project instructions and agent skills written for your codebase and version-controlled in your repository.
Orchestration of several agents: subtasks run in parallel, an independent review step, and verification against your tests before anything reaches the main branch.
Guardrails: permissions, automatic blocks on risky commands and guidelines for how secrets, source code and customer data may be handled, developed together with the team.
Review discipline and follow-up: how people review what the agents write, and which measures the team tracks to see whether the practice is working.

How it runs

A review of where you are today, workshops in your own repository on real tasks, a pilot workflow the team runs on its own, and a follow-up once the team has run the workflow for a while. On site around Falköping and across Skaraborg, or remotely anywhere in Sweden, in Swedish or English.

You get

Instructions, skills and workflows that live in your repository and belong to you, a pilot workflow in use, and a simple way to follow up on the effect.

Tools

We work with Claude Code and Codex. Instructions and workflows are written, as far as possible, to carry over between tools, so the practice is not tied to one vendor.

Frequently asked questions

Is this training or a consulting engagement?

Both. The team works in its own codebase from day one, and instructions, skills and workflows are built together with them. When the engagement ends, the team can maintain and extend the practice on its own.

What happens to our source code?

You choose the provider and the contract tier, and you decide which code and data may be sent there. Before any tool is used, we go through the provider's terms with you: what is sent to the service, how it may be used and what must be excluded. We work in your environment with the access you grant, and we do not access production data, secrets or customer data unless we have explicitly agreed it with you. Permissions and automatic blocks reduce the risk of mistakes in day-to-day work, but they do not replace your own routines and review.

Do you work with teams outside Skaraborg?

Yes. The engagement can be delivered entirely remotely, anywhere in Sweden, in Swedish or English.

Want AI to become part of how your team builds software?

Tell us about the team, the codebase and how you use AI today, and we will propose a plan.