What I Learned Onboarding Our AI Project Manager

Nityesh Agarwal describes building “Claudie,” Every Consulting’s AI project manager that saves 15 hours per week tracking client project status across email, Google Docs, Sheets, meeting transcripts, and calendar. The team is now trialing two more AI agents: Jean-Claude for sales pipeline and Claudette for visual design.

The road to a working agent was paved with previous iterations that had to be fired. Each failure revealed more about what it takes to make an agent a reliable coworker. Every’s CEO Dan Shipper calls this the “parallel organization chart” of AI colleagues, each with a name, manager, and real responsibilities.

The piece is paywalled after the setup, promising architectural workarounds for performance problems and the insight that Claudie’s first hard-coded task is reading her own employee handbook.

RDCO mapping: Strongest mirror of our own setup — named AI agents with defined roles, managed like employees. The “fire and rehire” iteration pattern matches our experience building the COO agent. The employee handbook concept is worth exploring for our own CLAUDE.md and SOUL.md architecture. Sponsor note: Rippling ad present (hiring guide).