Open Source · MIT Licensed

Developer
Infrastructure
for the Agentic Era

28 open-source projects across agent runtimes, accelerators, dev tools and governance. Drop-in by default. Rust where it matters. MIT-licensed throughout.

Browse all 28 projects View on GitHub
28
Open Source Projects
8
Product Categories
0
Config Required
MIT
Licensed

The agent stack is missing its infrastructure layer

Agents need isolated workspaces, durable memory, safe data access, payments, observability, and tooling that doesn't crumble at scale. We ship that layer — sometimes as drop-in Rust accelerators (LiteLLM, LangGraph, CrewAI), sometimes as new primitives (agentvfs, brat, mcp-pay, ormai), always open source.

The Problem

Python's GIL and single-threaded ORMs create latency spikes and OOM crashes under agent workloads.

Our Approach

We profile, isolate hot paths, and rewrite them in Rust using PyO3 for seamless Python interop.

The Result

Drop-in replacements. One import statement. Up to 700x faster with zero code changes.

Three pillars, one thesis

AI agents deserve the same caliber of tooling we built for the web era — runtime, performance, and developer surface. See the full 28-project portfolio on the projects page.

Agent Infrastructure

Orchestration, deployment, memory, payments, and safe data access — the runtime pieces an agentic stack needs.

Learn more
brat
Multi-agent harness for AI coding tools. Crash-safe state, parallel execution, one CLI.
fastagentic
The deployment layer for AI agents. Wrap PydanticAI, LangGraph, CrewAI, or LangChain and ship to production.
ormai
Give your AI agents database access without the risk. Policy-enforced ORM tools for Python and TypeScript.
mcp-pay
Payment awareness layer for MCP. A rail-agnostic manifest plus a reference 402 server.

Performance Accelerators

Rust-powered drop-in replacements for the Python libraries AI agents already depend on.

Learn more
fast-litellm
Drop-in Rust acceleration for LiteLLM. Faster connection pooling, rate limiting, and dramatically lower memory use.
fast-langgraph
Rust-backed checkpoint and state operations for LangGraph. Drop-in, zero config.
fast-crewai
Drop-in performance acceleration for CrewAI: faster serialization, tool execution, and database search.
fast-axolotl
Zero-config acceleration for Axolotl LLM fine-tuning. Faster streaming data loading and dedup.

Developer Tools & Build

Faster testing, faster builds, smarter CLIs — the developer surface for agent-era engineering.

Learn more
rjest
Drop-in replacement for Jest. A Rust daemon caches SWC transforms and pre-warms Node workers.
rpytest
Run your pytest suite faster. Change nothing.
rninja
Drop-in replacement for Ninja. Built-in caching and modern scheduling.
stkd
Stacked diffs, simplified. An open-source, Graphite-compatible CLI for GitHub and GitLab.

Zero-Effort Integration

Every accelerator is a drop-in replacement. No configuration, no code rewrites.

your_app.py 
# Step 1: Install
$ pip install fast-litellm

# Step 2: Add one import (before litellm)
import fast_litellm
import litellm

# Step 3: There is no step 3
# Your existing code is now 3.2x faster
For Investors

The picks-and-shovels play for the AI agent economy

Every AI company is building agents. Every agent needs infrastructure that works at scale. We're building the performance layer that makes production AI viable — open source first, with clear paths to enterprise monetization.

  • Open-source wedge — MIT licensed projects drive adoption at zero CAC
  • Deep moat — Rust+PyO3 expertise at the intersection of systems and AI
  • Platform expansion — From accelerators to orchestration to the full agent stack
  • Proven results — Measurable 3x-700x improvements across real-world workloads
Learn more about our vision
28
Public repositories
Rust
Core performance engine
Python
Seamless integration
Go
Workflow automation

Ready to accelerate your AI infrastructure?

All our projects are open source and MIT licensed. Start with a single import.

Read the Docs Browse on GitHub