A good fit
- software engineers, tech leads and engineering managers
- development, QA, platform and DevOps teams working together
- teams turning individual AI tool use into a shared development workflow
AI training for software engineers
Agentic coding training for software teams
AI agents are changing the entire software development lifecycle (SDLC). We teach teams to adapt planning, implementation, testing and release so they can deliver reliable software faster. Exercises use your company's coding tools, such as Claude Code, Codex or another coding harness.
How do we help your team adapt?
Map your development workflow to identify tasks agents can handle and decisions people own.
Practise a shared way of working, from clarifying a task to a tested, reviewed change prepared for release.
Build a setup the team can share, with project instructions, agent skills and automated checks.
Create a measurement plan for lead time, time waiting for review, rework and quality.
Who is it for?
Less suitable
- participants without basic programming knowledge
- teams seeking only general office AI training
How has the SDLC changed, and what will your team learn?
AI coding has moved from autocomplete to agents that inspect codebases, make changes and run tests across multiple steps. Developers spend more attention on defining work, directing agents and verifying results. Training puts this new division of work into practice throughout the software development lifecycle.
Requirements and planning
Delegated work needs a clear goal, acceptance criteria and technical constraints. Practise breaking tasks into manageable pieces, preparing context and reviewing an agent's plan. Clarify who owns architecture and product decisions.
Implementation with AI agents
Developers delegate tasks, follow the agent's work and intervene when needed. Practise delegation, debugging and small, verifiable changes in your team's chosen tool. Advanced exercises also cover parallel tasks and Git worktrees.
Harness engineering and shared team knowledge
Make useful working practices available to the whole team. Create AGENTS.md and CLAUDE.md project instructions, SKILL.md-based agent skills and automated checks. Practise context management, permissions, MCP connections and hooks supported by your chosen tool.
Testing and code review
As code is produced faster, testing and review can become bottlenecks. Turn acceptance criteria into tests, verify AI-assisted changes and decide what an agent can check and where human judgement is needed.
Release and production feedback
Agents can help beyond the pull request by preparing documentation and release notes or analysing CI failures. Practise checking release criteria, planning rollback and feeding production experience into the next development tasks.
Teamwork and measurable productivity
Coordinate developers, testers and leads around work with agents. Compare lead time, time waiting for review, rework, production defects and costs against the team's baseline. Use the results to decide which practices to keep.
Topics can form a focused workshop or a series of sessions. We use your team's tools; participants do not need to learn every coding harness. We agree on the syllabus, duration, access requirements and practice project in advance.
How do we tailor the programme?
01
Review the current development lifecycle
Trace how work moves from a request to production, where it waits and what causes rework. Choose training goals based on the team's AI experience and baseline measures.
02
Prepare the practice project
Choose an approved company repository or sample project and agree on access requirements and the expected result.
03
Practise across the lifecycle
Use a selected task to practise the new division of work from planning through release preparation. Review code, tests and the setup supporting agents together.
04
Plan team adoption and measurement
Document ownership, checkpoints and the practices to carry forward. Leave with a plan to trial them in daily work and measure their effect on speed, quality and effort.
Your codebase, agreed access
Exercises follow your company's data-handling and development rules. Alongside project instructions, we configure technical permissions and verification steps.
Approved repository or sample project
Accounts and subscriptions agreed in advance
Limited access and an isolated working environment
Tests and human code review
Frequently asked questions
How does this differ from an AI tool demo?
We work on the team's whole development process: preparing tasks, delegating to agents, testing, review and release. Claude Code, Codex and other tools are taught within that process, with shared ownership and verification practices.
How can we tell whether the team is more productive?
During training, create a measurement plan and record your baseline before adoption. Assess speed alongside quality and effort to see where the new working practices help and where they need adjustment. Results depend on the tasks and the team's way of working.
Does everyone need to learn both Claude Code and Codex?
No. Training centres on the tool your team uses or plans to adopt. Comparing tools can be a separate topic if needed.
Is advanced training available?
Yes. For teams already using AI tools, we focus on harness engineering, agent skills, integrations, parallel work and evaluating behaviour.
Can we use our own codebase?
Yes, with company approval and data-handling and access requirements agreed in advance. Otherwise, we choose a sample project suited to your stack and tasks.
How long does training take?
It depends on the topics and the team's experience. We design a focused workshop or a series of sessions and specify the duration in the training plan.