Software factory monitoring
AI turned your team into a software factory.
Who's watching the floor?
Teleos monitors the floor — every human, every bot, every AI agent committing to your repos, and the human who vouched for each change. Self-hosted on your own infrastructure: your code-health data never leaves your walls, and we never see it.
In n8n's public repo, 43% of the last 90 days of commits carried an AI tool signature. Could you name yours? See the read ↓
Public fishbowl
What the commit graph
actually shows.
n8n automates workflows on your infrastructure. Their open-source repo is a live factory floor. Teleos scans commit messages for tool signatures — mostly Co-Authored-By lines the tooling chose to write into git. Not seat logins. Not IDE telemetry. What landed in the repo.
Example report from public data · n8n-io/n8n · last 90 days
Commits
2,472
With detected signal
42.8%
1,059 commits
Seat → ship (latest)
38.3%
Of committers shipping code, how many left detected AI signal — not seat logins
AI tool mix
Which tools left a calling card in commit text — snapshot, not a trend
As of June 24, 2026
Weekly signal volume
How many detected tool signatures per week — trend over time
Seat → ship
You bought the seats. Are people using them on code that lands? Vendor dashboards measure logins. Teleos measures committers who left detected AI signal on shipped work.
15/39
Latest week · with signal / shipping
4-week rolling % of active committers with ≥1 detected AI tool signature in commit text. This is not your vendor's adoption rate — it's whether shipped code carries the calling card. No signal in git ≠ no AI; many workflows never write one.
Tool mix: point-in-time snapshot. Seat → ship: 4-week rolling % of active committers with ≥1 detected AI signal in commit text. Vendor adoption = logins; this = signal on shipped code. No signal in git ≠ no AI.
How it's built, not how it runs
Usage tools
Tell you whether people use the AI you bought. Per seat. On your laptop.
Runtime tools
Tell you how your software runs. They only see what you instrument.
Teleos
Tells you how it's built. Every change. Every author. Every voucher. Read from the one place all of it has to land.
Self-hosted
Your data.
Your infrastructure.
Full stop.
Teleos runs on your machine, your server, your cloud account. Repo names, engineer names, commit data — none of it touches our infrastructure. No DPA. No vendor questionnaire. No security review. There is nothing to review because there is no third party.
Pricing
Flat pricing.
Not per-seat.
Not per-contributor.
One operator gets the full product free, forever. When your team wants in, you buy it once — not a subscription, not per seat, not per contributor, never per agent. The per-seat model was invented before agents existed.
See pricing →Headless first
One command.
JSON out.
Your agents read it.
teleos scan owner/repo --json runs the
whole read on your machine and hands your agents — Claude Code, Cursor,
whatever you're running — the same record the dashboard shows.
No dashboard required. MCP server is next.
"I don't know who's in my code anymore."
Before AI, a good EM knew their team. Now the same repos get commits from agents, bots, and pipelines you didn't hire. The factory runs dark. You need a monitor.
"An agent merged it. Who answers for it?"
A robot can't be held accountable. Every change needs a human who vouched for it — the one who merged it, approved it, or owns that code. Teleos names them. Or tells you nobody did.
"In the absence of data, I'm managing by Slack status."
Perception-based management isn't lazy. It's the rational response to an information vacuum. Teleos fills the vacuum. Same judgment. Better instrument.
Every change has to land somewhere. Teleos watches where it lands.