Coinbase Rewrites Engineering Around AI Agents
Coinbase engineering leader Chintan Turakhia explains how his 1,000-plus-person organization is shifting from ticket-driven development to agent-first workflows. The playbook includes PR speedruns, automated feedback-to-PR pipelines, and measuring time from intent to production.
The important lesson is organizational, not merely technical: AI coding tools deliver outsized gains only when teams redesign workflows around them.
- –Engineers focus less on typing code and more on intent, orchestration, review, and verification
- –Coinbase uses visible experiments and hands-on leadership to drive adoption across a large engineering organization
- –Automated pipelines can turn spoken user feedback into structured tickets and proposed pull requests
- –Measuring time from idea to shipped product is more meaningful than counting AI-generated lines of code
- –The model raises a hard question for engineering leaders: if agents remove coding constraints, which human processes remain necessary?
DISCOVERED
2h ago
2026-08-18
PUBLISHED
2h ago
2026-08-18
RELEVANCE
AUTHOR
dakshgup