01—Case Study
Hermes
A personal AI employee that runs the back office.
Context
Running a solo consultancy means wearing every hat — sales, delivery, billing, scheduling. The admin work was eating into billable hours and creating gaps in client follow-up.
Problem
Email triage, calendar coordination, invoice generation, and follow-up reminders were all manual. Context switched constantly. Things slipped when focus was on client work.
What I built
Built Hermes as a coordinated agent system: an intake layer classifies incoming email, a scheduling agent syncs with calendar APIs, an invoicing workflow generates and tracks payments, and a follow-up agent monitors stale threads. Everything runs on a self-hosted stack with full audit logs.
Results
40+ hrs/mo
Admin time reduced
100%
Follow-up completion
99.9%
Stack uptime
Stack
- n8n
- OpenAI
- PostgreSQL
- Docker
- Tailscale