← back
Agents Don't Do Standups: Building the Post-Engineer Engineering Org — Mike Spitz, PFF
Takeaway
Redesign coordination around fast agent delivery, but judge success through customer outcomes and adopt the workflow gradually.
Summary
- A PFF pilot with two strong engineers reports 25x higher deployment frequency and 10x output under a blended ticket-and-complexity measure, with team-size caveats.
- The team delivered work in under two months against a four-month estimate and reported customer quality scores of 8.6 out of 10.
- Short product/design/engineering huddles replaced many Scrum ceremonies; agent interviews informed specs, lightweight design documents, tickets, and PRs.
- PR status automatically updated tickets, while rollout advice emphasized experienced engineers, gradual adoption, and initial experiments in noncritical systems.
engineering-organizationcoding-agentsdelivery-process
Original description
PFF ran a three-month case study: two engineers against a team of ten, same codebase, same customers. The two shipped five times a day. The ten shipped once every five days. Output measured by ticket complexity came out at 10x. Customer satisfaction went up, not down. Mike Spitz, their CTO, started with one reframe: stop asking how to help engineers go faster and ask how to make the agents faster instead. The talk covers what that reframe actually dismantled. Standups went away because tickets auto-update from PR state. Sprint planning went away because estimates are irrelevant when the bottleneck is no longer human. Code review got split: agents handle style and naming, engineers handle system design. The spec to lightweight design doc to auto-generated ticket to PR flow replaced most of the coordination overhead entirely. What survived is a short huddle every other day, a strong opinion about which engineers thrive in this setup, and a QA agent that spins up on staging after every merge and checks acceptance criteria against the ticket. Speaker info: https://x.com/mikespitz_uk / mike-spitz-89741243 Timestamps: 0:00 Introduction to the case study at PFF 1:47 The shift: optimizing for agent speed rather than engineer speed 2:28 Results: 25x increase in deployment frequency 3:16 Measuring success through ticket complexity and customer satisfaction 4:53 Dismantling Scrum and traditional development processes 5:58 The new development workflow (Spec → LDD → Ticket → PR) 6:51 Eliminating coordination overhead (no sprint planning or standups) 8:28 Best practices for implementation and team selection 10:09 Utilizing agents for deterministic tasks and code reviews 12:00 Viewing the engineering lifecycle as a factory 13:06 Automated QA and the future of self-healing systems 14:10 Where human oversight remains essential 14:48 Strategic advice for scaling AI-driven engineering