OLEV • YC-interviewed 2026
Canaries, a local agent, and audit reports for client IP.
Role
Co-Founder
Timeline
Mar 2026 – Present
Team
Founding team
Skills
Next.js
Python
FastAPI
PostgreSQL
Overview
Agencies using AI tools on client repos had no way to show what was actually read.
I co-founded OLEV: a local OS agent, GitHub canaries, coverage on specific repos, alerts, a people/device directory, and a compliance report. We log file paths and process names — not source code.
Detect
Canary tokens and an OS watcher record when an AI tool reads a protected file.
Scope
Coverage is per-repo, so tracking stays on client work instead of the whole machine.
Report
Those events export as a client-facing audit log.
Product
Platform
Coverage
Pick which client repositories get canaries and OS tracking. Internal repos stay off the list.
Alerts
A feed of reads: severity, actor (Cursor, Copilot, Claude Code, …), target file, and time.
Directory
People vs unidentified devices, with which AI tools showed up on each machine.
Outcome
Twelve pilots, and a YC interview in San Francisco.
12 pilots
Agencies using the product in real client work.
YC interview
In person, San Francisco.
Live
Shipped at tryolev.com.
Reflection
What I learned
Talk to the audience immediately.
The faster you are in front of the people who would buy, the less you build for an imaginary customer.
Make sure the problem is real.
A clever detection story is worthless if agencies are not actually losing deals over AI IP risk. Validate the pain before you scale the stack.
Move with agency.
Founding means deciding and shipping. Waiting for a perfect brief is how you miss the pilots and the interview.