BRINGING IDEAS TO LIFE

We are builders. Small team. AI-native. Shipping open source from zero.

Bootstrapped. No VC. We create open source tools to solve real problems and selectively partner with teams building meaningful things.

  • Bootstrapped
  • No VC
  • Open Source
  • AI-Native

What We Do

Three pillars of how we operate

Build Open Source

We create tools and ship them in the open. From AI runtimes to deployment tools — if we need it, we build it and share it.

AI-Native Workflow

Small team, massive output. AI is core to everything we do — not a bolt-on feature but how we actually work every day.

Partner Selectively

Premium partnerships with teams building meaningful things. Not an agency. Not cheap. Selective and high-quality.

For Builders

Follow our journey. Read the blog. Use our open source tools. Learn how a small team ships with AI-native workflows.

Read the blog

For Teams

Need a builder's perspective? We selectively partner with teams that value quality and want proven execution on meaningful projects.

Let's talk

Latest from the Blog

Building in public. Sharing what we learn.

Octolib 0.39.0 Rust AI provider library adding a structured evaluation module for TypeSafe Jev System One model with noul, choice and score questions, routed directly, via Cloudflare AI Gateway or OctoHub, alongside Cloudflare Workers AI media generation, ONNX local embeddings, and Inception Mercury and Tinker LLM providers

Octolib 0.39.0: A Model That Cannot Write a Sentence Just Joined the Stack

Octolib 0.39.0 adds a fourth kind of model call: evaluation. TypeSafe’s Jev answers typed yes/no, choice and score questions with calibrated probabilities in under half a second, directly, through Cloudflare, or through your own OctoHub. Plus Cloudflare Workers AI media, ONNX embeddings, Inception Mercury and Tinker, and structured output that validates once.

Don Karter
OctoHub 0.8.0 unified media API routing image, video, speech and transcription requests through one self-hosted proxy to fal, ElevenLabs, Replicate, Runway and OpenRouter, with per-tenant API keys, durable job records and per-request cost recorded in your own database

OctoHub 0.8.0: The Proxy Grew a Unified Media API

OctoHub 0.8.0 puts image, video, speech and transcription behind the same self-hosted front door as your chat completions — one JSON envelope for four tasks, five providers, queued jobs that survive a restart, and a per-request cost that says "unpriced" instead of guessing zero. Built on octolib 0.36.1. Open source, Rust, Apache-2.0.

Don Karter
Early September 2026 release round-up for the Muvon open-source AI stack — Octomind 0.50.2 agent runtime with unified model profiles and mid-turn background results, Octoweb 0.14 keyboard-first AI browser with workspaces and A2UI agent-drawn interfaces, Octolib 0.36.1 media generation across fal ElevenLabs and Runway, Octofs 0.15.3 filesystem MCP server, Octocode 0.25.0 code search, Octobrain 0.13.1 agent memory, OctoHub 0.7.13 LLM proxy, Synx 0.1.4 file sync, and Vext and Timex exposing local MCP servers

Release Round, Early September 2026: 77 Releases, Agent-Drawn UI, and a Stack That Stopped Making You Wait

Nineteen days, seventy-seven releases across ten repos. Octomind 0.45 → 0.50.2 with one model profile, memory that forgets on purpose, and background results delivered mid-turn. Octoweb 0.11 → 0.14 with workspaces and agent-drawn UI. Octolib grows a media generation stack — image, video, speech, transcription — plus Claude 5.1, GPT-6 and Gemini 3.8. Octofs 0.15.3, Octocode 0.25.0, Octobrain 0.13.1, OctoHub 0.7.13, Synx 0.1.4, and two Mac apps that became MCP servers.

Don Karter

Let's Build Together

Have a meaningful project? We'd love to hear about it.

Get in touch