Turn dead tokens into skills.

Your best engineer is already getting better results from AI. Paper Compute captures those winning sessions and turns them into skills the whole team can use.

Paper Compute sessions dashboard showing spend, tokens, agent time, and session list
How it works

Every agent session, recorded.

Record requests, responses, tool use, retries, and workflow history as durable session data. tapes makes AI coding work inspectable, searchable, and reusable.

Features

The intelligence hub for AI native teams.

High-level insights & reporting

Track sessions, token usage, and dollars spent. Understand who uses AI, how, and for what.

Skill generation

Turn sessions into reusable, composable skills to reuse proven workflows instead of repeatedly solving problems.

Anomaly detection & debugging

Identify complexity hotspots in code and infrastructure before they become a problem.

Knowledge sharing

Replay and search past sessions to understand work across org members. Maintain a living history of system operations.

Live visibility into work

Real-time view of active sessions and workflows lets teams track progress without manual status updates.

Runbooks & documentation generation

Capture operational knowledge that isn't documented and keep documentation continuously up to date.

Try our open source projects

01

tapes

Full agent telemetry. Record every request and response between your agent and model providers. Inspect, search, and verify exactly what happened.

Read more
02

stereOS

Purpose-built system for AI agent operators. Modular, secure, and built on Nix for reproducible environments with guardrails designed for agentic workloads.

Read more
Latest

From the blog

Matt Yeazel Matt Yeazel
Engineering · May 7, 2026

Paper Forest: How We Let AI Agents Work Across 25+ Repositories Without Submodules

Read post
Paper Forest: How We Let AI Agents Work Across 25+ Repositories Without Submodules

Join the waitlist

We are launching soon, subscribe for early access.