Who:
- Eddie Kim: Gusto CTO who scaled the payroll platform to $1B revenue before returning to hands-on coding
What Happened:
- Built Gusto Cofounder AI product with 3 engineers and 1 designer in 10 weeks
- Replaced Jira/standups with perma-Zoom and trash-can PR method (delete after review)
- Non-technical designer shipped production code using Claude Code in eval-first workflow
Why It Matters:
- Challenges enterprise dogma that AI products require massive teams and documentation
- Proves eval-driven development (fixing real customer bugs first) beats spec-driven planning
- Designer-to-engineer pipeline via Claude Code could reshape talent acquisition
ARM Impact:
- Tab Hopper (Stage 1 (Tab Hopper)): Proves AI can compress prototyping-to-production cycles by 10x
- AI Sprinkler (Stage 3 (AI Sprinkler)): Designer coding via Claude shows non-engineers can own full-stack features
- ARM (Stage 4 (Autonomous Revenue Master)): Trash-can PR method eliminates bureaucratic drag from version control
What to Watch:
- Whether Gusto open-sources their Claude Code eval templates for payroll domain
- If other Fortune 500 CTOs replicate this model for internal AI tools
- Adoption of perma-Zoom as always-on alternative to async comms tools