Agentic Software Development
From shared context to verified delivery.
Most teams already use AI every day, as a chat window and an autocomplete. Agents that read, change and run code are a different tool: they need a method, and someone who stays accountable for what ships. This course teaches both.
What your team
will be able to do.
What each participant can do on their own by the end of the core course.
- 01
Turn a business request into a bounded, verifiable task.
- 02
Choose tools and ways of working with quality, cost and company constraints in mind.
- 03
Build a second brain for the team: a shared, versioned knowledge base that people and agents can both query.
- 04
Take a task from analysis to review while keeping ownership of the decisions.
- 05
Build a reusable skill and check that it actually helps.
- 06
Recognise errors and risks, and know when to step in directly.
Who it is for,
and how it runs.
Explanations, live demonstrations and hands-on labs, around one project that grows from the first module to the final lab.
Who it is for
Developers, tech leads and consultants who write code every day. No previous experience with agentic coding is needed.
Before the first session
Working knowledge of Git, the terminal and automated tests. Licences, access and tooling are checked beforehand, not during class time.
On-site or remote
One team at a time, at your office or over video, with exercises adapted to your stack and conventions.
Your AI provider
Runs on the tools your company has already approved, such as Claude Code, Codex or Gemini CLI. The method does not depend on the vendor.
A written specification in the repository before any code, agents that implement against it, and a person who reviews every change.
It is how we deliver client projects, supported by Seshat, our internal toolkit for spec-driven agentic development.
See it on a client project