Agentic Engineering Bootcamp
Ship real ticketswith coding agents.
A hands-on bootcamp in your own codebase. Your engineers use Claude Code and Codex on real tickets, learn where agents fail, and keep the workflows, prompts and guardrails they build.
Why it matters
Coding agents are fast. Your team needs to know where they fail.
- Where
- In your codebase, on tickets from your own backlog.
- Tools
- Claude Code and Codex by default, or the tools your team already has approved.
- Access
- The permissions and environments your security team approves.
- Who
- Engineers at any level. No AI experience needed.
Run by forward deployed engineers from an OpenAI partner, with 5 certified engineers.
How it works
Learn on real work, not toy examples.
Every session uses your codebase and your backlog, so what the team learns works on Monday.
- Set up
Prepare the codebase
Agree the repositories, tickets and access. Set up the agent tools with the permissions your security team approves.
- Ship
Pair on real tickets
Engineers work real tickets with coding agents, from planning to tests to review. We coach in the session, not from slides.
- Break
See where agents fail
Work through the failure cases on purpose: wrong assumptions, silent test gaps, unsafe changes and leaked secrets.
- Keep
Make it the team's way of working
Write the team's agent instructions, review checklist and guardrails into the repo. The bootcamp runs over two days for up to 12 engineers.
Scope
What the team practises.
The bootcamp covers the full loop, not only prompting.
- Plan
Scoping work for an agent
Turning a ticket into a task an agent can do well, with clear checks for done.
- Build
Agent-driven implementation
Using Claude Code and Codex to write, refactor and test code in a large, real codebase.
- Review
Reviewing agent-written code
What to check, what to distrust and how to keep pull requests small enough to review.
- Secure
Safe agent use
Permissions, secrets, sandboxing and the changes an agent must never make alone.
What you take away
Deliverables your team keeps.
Everything we produce belongs to your team and lives in your accounts and repositories.
Yours to keepEverything we make is yours.
Every output stays in your accounts and repositories when we leave.
- 01 / Shipped
Real work merged
Tickets from your own backlog, shipped and reviewed during the bootcamp.
- 02 / Workflows
Agent setup in your repo
Instruction files, commands and checks the whole team uses after we leave.
- 03 / Guardrails
Rules for agent-written code
A review checklist and security rules your team agreed and can enforce.
- 04 / Baseline
A before-and-after measure
The measure the team chose, for example cycle time or review time, taken before and after.
Sample output
What stays in your repo.
The bootcamp ends with files in version control, not slides in a shared drive.
AGENTS.mdTeam instructions the agents read on every task.claude/commands/Shared commands for planning, testing and reviewdocs/agent-review-checklist.mdWhat reviewers check in agent-written code.github/workflows/agent-checks.ymlAutomated checks on every agent pull requestdocs/bootcamp-baseline.mdThe before-and-after measure the team chose
Illustrative format. File names and contents depend on your repository, your tools and the rules your team agrees.
Before we start.
Discuss anything specific to your team or environment.
Claude Code and Codex by default. We can work with the tools your team already has approved.
No. The bootcamp starts from your codebase, not from AI theory. Experienced users go deeper on review and safety.
We use the permissions and environments your security team approves. We agree the repositories and data in scope before we start.
AI Training teaches teams to build AI features. This bootcamp teaches engineers to use coding agents to build any software faster and safely.
The next step
Put agents to work on your backlog.
Request a technical callWe’ll confirm fit and the scope in writing before any work begins.
