Agentic coding as a team
Two days with your engineering team: introduce Claude Code and Codex so the code passes review. On your own repository, with real tickets and shared rules that apply from the Monday after.
Training for software teams · 2 days · on site or remote
The problem
Everyone uses AI differently. Nobody knows what good looks like.
Claude Code, Codex, Copilot or Cursor have arrived in the team, but without a shared way of working. Some build agentically, some stay sceptical, product waits for results and wonders about the reviews. The training gets the whole team to one table, names the problems and builds the way of working where it has to hold: in your repository.
- Engineering teams of 4 to 12 people, optionally with product
- Teams that already use AI coding tools, but without shared rules
- Prerequisite: Git, a real repository, access to at least one coding agent
How it works
Two days on your own repo
Every block ends with something your team keeps using.
Day 1 morning: how the tools think
Context, planning, specs and the limits of an agent. Live on your codebase, not on slides.
Day 1 afternoon: real tickets
Everyone works through a real ticket with the agent. We compare approaches and write down what worked.
Day 2 morning: rules and review
Shared rules for prompts, commits and reviews. Gates that catch what the agent gets wrong.
Day 2 afternoon: your playbook
Repository setup, team playbook and the first three things you change on Monday. Handed over, not promised.
What you take away
- A team playbook for AI-assisted development in your own repository
- Shared rules for prompts, commits and reviews
- Hands-on experience from real tickets, not demo projects
- A clear view of what to automate next and what to leave alone
Why me
Introduced in a real team, not on slides
As Head of Engineering at Seobility I introduced AI workflows during running operations. I have taught at iba University of Cooperative Education and given seminars at GFU and other academies. I work with Claude Code and Codex myself every day.
- Seobility: AI workflows in engineering, Head of Engineering
- Teaching: iba (until 2025), seminars at GFU and academies
- AI Nights: a community of more than 1,000 participants
Good to know
Questions about the training
Which tools do we use?
Claude Code and Codex are the core. If your team works with Copilot or Cursor, we include them. The rules we agree on apply to every tool.
On site or remote?
Both work. On site is better for the first day, remote works well for follow-ups. We decide based on your team.
What if the team is bigger than twelve?
Then we split into two groups or run the three-day Factory Workshop, which is designed for whole departments including product.
In German or English?
Both, also mixed, depending on how your team is set up.
Ready for the training?
Let's talk about your team and your repository. Free and non-binding, 30 minutes is enough for a first assessment.
Book a call