Thanks for attending my recent talk. Below you'll find some resources that I think may be helpful as it relates to scaling Code Agent use in your engineering team.
Survey: understanding AI usage in engineering teams
I'm currently doing some research on how engineering teams are using AI in practice today. If you have a few minutes to fill out a quick survey I would really appreciate it. I will share the results back to the community when they are ready.
Talk slides and notes
You can view the slides from my talk below, or alternatively visit the site for the large version.
Slides from my talk at Web Directions AI Engineering 2025, including notes, are available here for download (PDF 47MB).
Resources
These are some resources that I've found that have been genuinely helpful. If you come across something you think should be on here then please reach out to me and let me know
Coding agents (non exhaustive)
- Claude Code
- Open AI Codex
- Gemini CLI
- Github Copilot Coding Agent
- Atlassian Rovo Dev agent
- Cursor IDE (in agent mode)
- Aider
Documentation and context for agents
- Focus on your README file first and foremost, even for agents
- Have one agents file, even with multiple tools working on the repo - Agents.md
- How documentation supports better pair programming with code agents and assistants
Monorepo strategies and structure
- Monorepo set up in 2025
- Structural approaches and examples for monorepo setup
Agent Parallelisation
- Testing parallelism and sub tasking agents in Claude Code
- Methods of running multiple agentsin parallel locally and cloud based
- Practical breakdown of orchestrating multiple tasks in parallel using Claude Code
Guardrails, review and workflow automation
- How to use LLMs in GH actions to validate changes or add new workflows
- Using LLMs as part of basic CLI commands (uses Simon Willison’s LLM project)
- Similar approach with LLM tool to build meaningful commit messages on the fly
- Great view of defence in depth and multiple ways to set it up at local and remote levels to create good guardrails
- Research paper suggesting ways to build in real-time analysis and defence for coding agents to mitigate risk (LlamaFirewall)
- Recommendation checklists on validation flows needed for agent produced code (especially vibe coding scenarios but applies more generally too)
- Silly but fun way to make code reviewing fun by generating images
Good practices and documentation (from toolmakers)
- Best practice approaches on using Claude Code
- Working with Open AI Codex
- Gemini CLI live documentation
- Gemini CLI code lab
- Tips for working with Aider
- Using Github Copilot agent to work on tasks
Other miscellaneous references
- A full, free book on Agentic Design Patterns by Antonio Gulli (424 pages on how to build with agents)
- Good article about coding with cursor in assistive and agent modes
- Useful walk through of using all of Github’s tooling across multiple workflows to show how different parts fit together
Where to next?
If you got this far then thanks for reading and a reminder about the survey I'm running to understand how teams are using AI in practice today. It only take a few minutes and I would really appreciate it.
The home page will give you featured articles and recent posts.
You can also check out the full list of articles that have ever been published.