Agentic Engineering: Working With AI, Not Just Using It
summary
Coding agents move from autocomplete to autonomous collaborators that can execute tasks, touch files and produce pull requests, and teams report roughly 30% time savings when they know what to hand off. Success depends on concrete practices: the research–plan–implement loop, mode-separated workflows like ask/code/architect, persisting key state in an agents.md and using Git and frequent commits for reviews. Watch for pitfalls: MCP servers and too much context — quality degrades past about 50% context — and tune autonomy vs human approval.
AI-written summary · about
chapters (16)
- 0:00Agentic engineering intro
- 1:04AI's rapid evolution
- 3:09Mental model for AI agents
- 4:32Why context engineering matters
- 7:21Managing context
- 9:07Lessons from junior engineers
- 11:32Research plan loop
- 13:38Why research and planning matter
- 16:27Implementing changes
- 17:51Customizing agent modes
- 19:36Using agents.md and skills.md
- 21:00Power user tips
- 22:47Expanding beyond IDEs
- 23:50Managing MCP servers
- 25:33Reviewing agent work
- 26:17Next steps
related talks
29mPairing with the machine: How AI Agents are rewriting the rules of software development
26mAI Agents as Your Co-Developer
46mAgentic Engineering: From Writing Code to Reviewing It
28mAgentic AI
49mFrom autocomplete to agents: AI coding assistance state of play
01h 00mKenote: AI Didn’t Replace You. It Promoted You!
22mAgentic AI for Engineering Leaders | Strategy & Execution Guide
27mEngineering in the Age of AI: Making AI Work for Us, Not Replace Us