The Stack Overflow Podcast Craft and quality beat speed and scale, with or without agents
24 snips
Oct 28, 2025 Tom Moor, Head of Engineering at Linear, dives into the dynamic world of AI agents in software development. He shares insights on how context enhances agent performance and discusses the balance between AI's hype and practical adoption. Tom highlights how junior developers can leverage AI to improve their skills while collaborating effectively with these tools. He also touches on the importance of creating workflows that allow for iterative feedback and quality improvement, showcasing Linear's innovative integration of agents within team environments.
AI Snips
Chapters
Transcript
Episode notes
Linear As An Agent-Ready Issue Tracker
- Tom Moor describes Linear as an issue tracker that holds work, context, and assignments for teams building software.
- He explains agents get assigned work in Linear and return PRs or answers as team members do.
Give Agents Small, Precise Specs
- Give agents precise constraints and small specs to reduce mistakes and improve output.
- Include names, patterns, and file locations so the agent can pattern-match and produce usable code.
Few Bad Runs Undermine Adoption
- A few bad experiences make engineers abandon agents even if small prompt fixes would help.
- Investing 30 extra seconds refining context can often save many minutes of manual work later.