# PeepSick Labs

> Independent AI infrastructure studio building agent tooling in public.
> Pre-incorporation, bootstrapped, pre-revenue.

PeepSick Labs builds the plumbing AI agents need and mostly do not have — a
portable way to ship skills, one API surface across model providers, a runtime
for making several agents work as one — plus the applications that prove those
pieces hold up in a real business. Each piece is small, does one thing, and is
useful on its own.

Three products are infrastructure (USB, LeoSIS, Foundry) and three are built on
it (ModelForce, Mühendis, İzMercek).

Available today: USB, ModelForce, Mühendis.
In development, not yet available: LeoSIS, Foundry, İzMercek.

## Products — available

- [USB — Universal Skill Bridge](https://peepsicklabs.com/usb/): One portable skill format installed natively into 16 agent runtimes with sha256-verified installers. 529 skills in the catalog. Live in beta, free and open source.
- [ModelForce](https://peepsicklabs.com/modelforce/): A runtime and provider infrastructure for self-hosted text-to-speech. One API for Piper, Kokoro and XTTS — local inference, CLI/SDK/REST/Web UI. Open source, MPL 2.0, early development.
- [Mühendis](https://peepsicklabs.com/muhendis/): An AI operating system for mechanical contracting — quantity takeoff, pricing, procurement, stock, accounting and finance, with a multi-agent layer and RAG over regulations and material libraries. Live, in Turkish, built on LeoSIS.

## Products — in development

- [LeoSIS](https://peepsicklabs.com/leosis/): OpenAI-compatible LLM provider layer — one API surface across model providers. Not yet available.
- [Foundry](https://peepsicklabs.com/foundry/): Multi-agent orchestration runtime for agents that share state and hand work to each other. Not yet available.
- [İzMercek](https://peepsicklabs.com/izmercek/): AI visibility measurement and knowledge publishing — compiles canonical brand knowledge into AI-readable artifacts and publishes them to the brand's own site. Not yet available.

## Capabilities

- Natural language compilation: Convert human intent into reliable, executable artifacts.
- Model-independent AI infrastructure: Build AI systems that are portable across providers and runtimes.
- Evidence-driven verification: Ensure outputs are explainable, reproducible, and backed by verifiable evidence.
- Distributed AI orchestration: Coordinate models, tools, and services through a unified execution layer.
- Portable capability packaging: Deploy reusable AI skills across different environments without vendor lock-in.

## About

- [About PeepSick Labs](https://peepsicklabs.com/about/): What the studio builds, how it is run, and what "building in public" means in practice.
- [Contact](https://peepsicklabs.com/contact/): info@peepsickai.com for general enquiries; GitHub issues for technical problems.

## Code and packages

- [github.com/PeepSick](https://github.com/PeepSick): All public repositories.
- [github.com/PeepSick/usb](https://github.com/PeepSick/usb): USB source and issue tracker.
- [github.com/PeepSick/ModelForce](https://github.com/PeepSick/ModelForce): ModelForce source.
- [@peepsick/usb-cli](https://www.npmjs.com/package/@peepsick/usb-cli): USB command line tool.
- [@peepsick/usb-sdk](https://www.npmjs.com/package/@peepsick/usb-sdk): SDK for building on USB.
- [usb.peepsicklabs.com](https://usb.peepsicklabs.com): USB's own site.
- [muhendis.birseyleryap.com](https://muhendis.birseyleryap.com/): Mühendis, live.

## Citation

PeepSick Labs — independent AI infrastructure studio, building agent tooling
in public. Source: https://peepsicklabs.com

## Optional

- [llms-full.txt](https://peepsicklabs.com/llms-full.txt): Extended version with per-product detail and FAQ.