Part of Red River WestPart of Red River WestLearn more →

>commit portfolio

21 commercial open-source companies — 10current >commit Fund I investments and 11 companies the team backed before the fund.

>commit Fund I (current portfolio)

  • Macrodata Labs — Every strong model starts with great data. Macrodata Labs builds training-data infrastructure for physical AI. Its open-source framework, Refiner, turns raw multimodal robotics data (trajectories, camera feeds, audio, language) into high-quality training datasets, running locally for development and scaling to an elastic serverless cloud with a single command. Built by the team behind FineWeb, the largest open LLM pre-training datasets.
  • Paradev — Run an army of coding agents. Paradev (formerly Superset) is an agentic IDE for orchestrating coding agents in parallel. One workspace runs Claude Code, Codex, or any other coding agent on your own subscription, each in an isolated Git worktree so they never step on each other’s files, with built-in PR review, remote workspaces, and scheduled automations. Built by three former CTOs of YC-backed companies.
  • UMA — Intelligent robots that enhance quality of life for everyone. UMA builds humanoid robots that combine human-level dexterity with a deep understanding of the physical world. The team blends AI research from DeepMind, Tesla Autopilot, and HuggingFace with decades of humanoid-robotics hardware experience to ship general-purpose machines that can take on everyday physical tasks.
  • ZML — Production inference stack decoupling AI from proprietary hardware. ZML is a production inference stack that decouples AI workloads from proprietary hardware. From a single codebase it compiles large language models — Llama, Gemma, Qwen, Mistral — to run at peak speed across NVIDIA, AMD, Google TPU, Intel, and Apple Metal, letting teams mix accelerators, break vendor lock-in, and cut inference cost and energy. Its LLMD inference server, built on Zig, MLIR, and OpenXLA, launched free in 2026.
  • Stealth — AI agent runtime control.
  • Stealth — AI research infrastructure.
  • Stealth — Open-source backup platform.
  • Stealth — Registry for AI agent skills.
  • Stealth — Specifications framework.
  • Stealth — Verifiable trusted execution environment.

Pre-commit (companies the team backed before the fund)

  • Mastra — TypeScript AI agent framework. Mastra is an opinionated TypeScript framework for building agents, workflows, RAG pipelines, and evals. It ships the primitives (memory, tools, tracing, voice) under a single API so teams can go from a notebook prototype to a production agent without reassembling the stack each time.
  • Twenty — Enterprise CRM at AI Speed. Twenty is an open-source CRM platform: a developer-first extensibility toolkit for composing customer and internal apps without vendor lock-in. It ships production-grade building blocks (data model, permissions, workflows, authentication), runs locally like normal software, and is AI-native via an MCP server so agents can iterate on the schema, layouts, and automation directly.
  • Better Auth — Auth that lives inside your app. Better Auth is a framework-agnostic TypeScript authentication library that lives inside your app rather than behind a vendor API. It ships session management, social sign-on, two-factor, and organizations out of the box, and a plugin system that lets teams extend the auth layer instead of outgrowing it.
  • Atuin — Making your terminal magical. Atuin replaces your shell history with a searchable SQLite database, then syncs it between machines with end-to-end encryption. It records the context around every command — exit code, duration, directory, session — so history becomes something you can actually query, and ships a self-hostable server for anyone who would rather keep their own data.
  • Sourcebot — The Code Understanding Platform. Sourcebot is a self-hosted code understanding platform. It indexes every repository in an organization, answers natural-language questions across them, and powers AI-assisted search so teams can navigate large, multi-service codebases without rebuilding context every time.
  • pyannote — Speaker intelligence for developers. pyannote turns real-world audio into structured, programmable intelligence — speaker diarization, voice identification, overlap detection, and real-time streaming — delivered through developer APIs and built on the widely-used pyannote.audio open-source toolkit.
  • Pangolin — The All-in-One Remote Access Hub. Pangolin is an identity-aware remote-access hub built for IT/OT, IoT, and engineering: a self-hosted reverse proxy and tunneled VPN that gives users secure, SSO-backed access to applications and infrastructure, with context-aware rules and a unified control plane.
  • White Circle — Control your AI. White Circle is a control layer for AI applications — stress-testing models for jailbreaks and hallucinations, enforcing low-latency safeguards at runtime, and auto-patching vulnerabilities — so teams can ship LLM features with confidence.
  • Keep — The Open-Source AIOps Platform. Keep is an open-source AIOps platform — a Swiss-knife for managing alerts and events at scale. It correlates noisy signals from monitoring tools, runs automated workflows across them, and ships with AI-powered deduplication so on-call teams see fewer pages and more signal.
  • PandasAI — AI Dashboard for Business Intelligence. PandasAI makes data analysis conversational. It lets teams chat with SQL databases, datalakes, and local files (CSV, Parquet) using LLMs and RAG, delivering business-intelligence answers without writing SQL or Python.
  • Graphcore — Accelerating machine learning in the cloud. Graphcore builds Intelligence Processing Units (IPUs) — a new class of silicon purpose-built for machine intelligence — alongside an open toolchain that lets developers run PyTorch, TensorFlow, and custom models on IPU-accelerated infrastructure in the cloud.