One user made 24,000 Copilot requests in 24h: how firms audit 580 AI seats nobody tracks
alex_verem · x · 2026-10-17
An engineering leader at a PE-backed financial firm admitted that after rolling GitHub Copilot out to everyone, they had "completely lost track" from an audit perspective—invoices only count seats, not usage.
Before an audit, he wrote a script pulling 28 days of usage across 580 users. It surfaced two problems invisible on the invoice:
- One person made 24,000 requests in 24 hours via the command line—one every 3.6 seconds—in violation of usage rules.
- Teams expected to use AI had barely made any requests.
The post lays out five audit questions for any portfolio company: who used their seat in the last 7 days, who hasn't touched it in 28, which accounts exceed human-possible request rates, and more. The takeaway: seat-based AI licensing needs active usage auditing, or spend disappears into unused licenses.
More from coding & agent
- Having your agent read 8,000-word articles may be the future of serious online writing — danshipper · 2026-10-17
- Tencent open-sources Hy-Memory 2.0 for cross-tool agent memory sharing — aigclink · 2026-10-17
- Approval flows should follow the risk, not the org chart — usehive · 2026-10-17
- OpenDesign Motion: An Open-Source Alternative to Claude Motion for Video — petewoodbridge · 2026-10-17
- Durable PTY: Open-Source Tool Keeps AI Coding Agents Alive Across Crashes and Reboots — jasonkneen · 2026-10-17
- Deer forest demo vibe-coded in 1 hour with Tripo API and Claude — Scobleizer · 2026-10-17