Agentic Engineering with Claude Code

A practical, full-day course for developers on how to use Claude Code effectively in software development.

Claude Code is the main tool, but most methods also apply to OpenAI Codex and other agentic coding tools.

As AI agents take on more of the coding, the bottleneck is moving from writing code to defining the work, feeding agents the right context and verifying the outcome. The most capable models also require a different approach to prompting and context management. The goal is not to provide more context, but to provide the right context at the right time and build reliable checks around the work.

The course covers how software work can be organised around AI agents. This includes defining the task, providing context, setting constraints and choosing how results are checked. Agents can then explore, plan, implement, test and review the work.

Who should attend?

Software developers, tech leads, engineering managers and technical founders who want hands-on experience using AI agents in their development workflow.

You should be comfortable reading, editing and running code in a Git repository. Previous Claude Code experience is useful, but not required.

What you will learn

• How to structure work across exploration, planning, implementation and verification
• How prompting and context management change when working with the most capable models
• Context engineering with CLAUDE.md, memory, compaction and just-in-time context
• Feedback and verification loops, including software factory patterns that connect implementation, testing and independent review
• Background agents, isolated Git worktrees and parallel workflows
• Subagents and multi-agent harness for focused research, implementation and review
• Compound engineering and turning lessons into instructions, tests, skills and tools
• Permission rules and how to inspect, steer and resume longer-running work
• Practical tips and tricks for choosing models, prompting effectively and handling common workflow problems

How we will work

This is a hands-on course built around development tasks in a prepared Git repository. You will work directly with Claude Code and build a reusable setup with project instructions, verification checks, a skill and a workflow for running and reviewing agent work.

The exercises cover interactive sessions, background work, parallel agents, isolated worktrees and capturing lessons for the next task. Along the way, we will share practical tips and tricks, useful shortcuts and patterns drawn from day-to-day work with coding agents. You can use Claude Code in the terminal, desktop app or an IDE such as VS Code or Cursor.

Preparation

• A Claude plan with access to Claude Code
• Laptop with Claude Code installed in your preferred interface, either the terminal, desktop app or the plugin for VS Code or Cursor
• A development environment with Git
• A prepared repository will be provided. You can also use your own codebase where appropriate

Your lecturers

Aleksander Stensby is a Claude Code Community Ambassador and hosts the Oslo meetup. As the former CEO of Klaveness Digital, he built and scaled a B2B SaaS platform. He has co-founded several startups and speaks at NDC AI and other conferences.

Simon Sebergsen is the former CPO and CCO at Klaveness Digital, with a background in finance, strategy, product and growth. He works with Claude Code and agentic workflows across software, content and operations.

About GritAI Studio

GritAI Studio is an AI studio based in Oslo. We host the official Claude Code Meetup at Mesh and are part of Anthropic's Claude Community Ambassador Program.

We help companies adopt AI through training, workshops and consulting across Norway and Europe.

Learn more at gritai.studio / YouTube / LinkedIn