Field Notes

GitHub's Burke Holland outlines a repeatable 8-step Copilot workflow (prototype, plan, autopilot implement, human review, rubber duck review) arguing that mastering the core agent harness—not chasing new tools/skills/MCPs—drives most productivity gains.

A framework (packaged as an AI PM OS skill) argues product teams fail by answering the first question they encounter rather than treating question-selection as its own discrete task, using wide-exploration techniques and Warren Berger's rewrite methods to surface higher-value questions before ranking and committing to an action.

A dense list of specific outbound sales tactics covering email deliverability hygiene, enrichment/data quality practices, and high-signal trigger events for B2B prospecting, from @fin465.

A framework distinguishing three problem-solving question types—Why (causes), What (work breakdown), and How (actions)—to prevent teams from conflating diagnosis, planning, and solutioning, packaged as an AI-assisted skill in AI PM OS.

A framework distinguishing three separate layers in production agent systems: harness engineering (environment/tools/state), loop engineering (feedback/retry cycles), and graph engineering (explicit workflow topology). Includes diagnostic rules, common mistakes, and a production checklist, plus a reference to OpenAI's Forward Deployed Engineer (FDE) talk on production AI deployment.