Rust + Python
Drop-in Rust acceleration for LiteLLM. Faster connection pooling, rate limiting, and dramatically lower memory use.
36 open-source products · MIT licensed · UK studio
The infrastructure lab
for the
agentic era
Neul Labs ships a family of 36 open-source tools for AI builders — Rust accelerators for the Python libraries you already use, agent infrastructure and runtimes, agent SDKs, and faster dev tooling. Drop-in by default. Rust where it matters. MIT throughout.
36
Open-source products
9
Product categories
0
Config required
MIT
Licensed throughout
What is Neul Labs
An open-source lab building the layer the agent stack is missing
Every AI team is building agents. Those agents need isolated workspaces, durable memory, safe data access, payments, governance and tooling that doesn't crumble at scale — and the Python libraries underneath them hit latency, memory and cost walls under real load.
Neul Labs ships that missing layer as 36 focused, MIT-licensed products. Sometimes it's a drop-in Rust accelerator for a library you already use; sometimes it's a new primitive — an agent sandbox, a policy gate, a memory layer, a native agent SDK. Each one does one thing well and lives at its own subdomain.
drop_in.py
# One import, no rewrite — your code just gets faster
import fast_litellm # before litellm
import litellm
# Or reach for a new primitive when you need one
from memorg import Memory # durable memory
from ormai import guard # safe DB accessThe portfolio
36 open-source products, one thesis
Each product lives at its own subdomain, ships its own docs, and encodes one sharp idea. Filter by what you're building.
Rust + Python
Rust-backed checkpoint and state operations for LangGraph. Drop-in, zero config, 737× faster checkpoint serialization.
Rust + Python
Drop-in performance acceleration for CrewAI: faster serialization, tool execution, and database search.
Rust + Python
High-performance Rust extension for Axolotl LLM fine-tuning. 77× faster Parquet streaming, parallel dedup, no OOM.
Rust + Python
A drop-in LangGraph replacement on a Rust engine. Same graph API, native performance underneath.
Rust
Multi-agent harness for AI coding tools. Crash-safe state, parallel execution, one CLI.
Python
The deployment layer for AI agents. Wrap PydanticAI, LangGraph, CrewAI, or LangChain and ship to production.
Python
Brokerless Python task queue. No Redis. No RabbitMQ. Just 2–3 Python processes.
Go
Zero-infra, single-binary Go workflow automation runtime. The n8n alternative without the ops.
Rust
One CLI for every AI coding agent — with isolated profiles, provider switching, and cost tracking.
Python
Give your AI agents database access without the risk. Policy-enforced ORM tools for Python and TypeScript.
Rust
Payment awareness layer for MCP. A rail-agnostic manifest plus a reference HTTP 402 server.
Python
External memory layer for LLM apps. Hierarchical SQLite + vector search, deterministic recall, MCP server.
Rust
Workspace runtime and execution boundary for AI agents. Isolated vaults, instant forks, rollback points.
TypeScript
Self-hosted, multi-channel AI agent gateway with an OS-like, process-isolated architecture.
TypeScript
The multi-tenant AI gateway your platform can actually share. Enterprise tenancy for OpenClaw.
Rust
A Rust implementation of OpenClaw. Event-sourced agent runtime with capability-based sandboxing.
Dart
The mobile-native AI agent SDK for Flutter and Dart. On-device and cloud models, subagents, skills, MCP.
TypeScript
The in-browser AI agent SDK. Run agents client-side, in the page, with no server round-trip.
TypeScript
The agent SDK where humans drive the state. Human-in-the-loop agents by construction.
Zig
The systems-language agent SDK. A 960 KB Zig binary agent with a C ABI and Python/TS/Go SDKs.
Python
AI assistant for the buy-side fixed-income desk — and a reference build of the FINOS AI Governance Framework.
Rust
Drop-in replacement for Jest. A Rust daemon caches SWC transforms and pre-warms Node workers.
Rust
Run your pytest suite faster. Change nothing.
Rust
Make large Git repositories feel instant.
Rust
The issue tracker that lives in your repo. Built for AI agents. Works for humans.
Rust
Intelligent CLI assistance from Claude Code agents. Generates SKILL.md files and serves them at your Tab key.
Rust
Drop-in replacement for Ninja. Built-in caching and modern scheduling cut warm builds 2–5×.
Rust
10–100× faster, Rust-based drop-in replacement for Homebrew.
Rust
Stacked diffs, simplified. An open-source, Graphite-compatible CLI for GitHub and GitLab.
Rust
curl that just works. Drop-in replacement with automatic anti-bot bypass.
Rust
wget, but it works everywhere. Drop-in replacement that bypasses bot protection.
TypeScript
A semantic filesystem that turns Local / S3 / GCS / Azure / SharePoint / Drive / Jira storage into an AI-searchable knowledge base.
Java
The open-source EU + UK compliance plane for Google ADK. Ship AI agents your regulator accepts.
Python
The policy chokepoint for finance AI agents.
Dart
Build personal AI agents on your phone. Describe a task in plain language; Ukkin runs it in the background.
No products in this family.
What we build
Four families across the agent stack. Every product fits into one of them.
Rust Accelerators
5 products
Drop-in Rust replacements that speed up the Python AI libraries you already use.
Agent Infra & Runtime
15 products
The runtime layer agents need: orchestration, memory, sandboxes, data access, payments, governance.
Agent SDKs
6 products
Frameworks for building agents — on the phone, in the browser, in a C ABI, or on a governed desk.
Find a tool
If you need this, reach for that
A decision table across the whole portfolio. Match your problem to the product built for it.
| If you need to… | Reach for | Family | Stack |
|---|---|---|---|
| speed up LiteLLM without changing your code | fast-litellm | Rust Accelerators | Rust + Python |
| accelerate LangGraph checkpoints and state | fast-langgraph | Rust Accelerators | Rust + Python |
| make an existing CrewAI crew faster | fast-crewai | Rust Accelerators | Rust + Python |
| fix Axolotl data-loading RAM/OOM at scale | fast-axolotl | Rust Accelerators | Rust + Python |
| run LangGraph-style agents on a Rust engine | agentmatic | Rust Accelerators | Rust + Python |
| run several coding agents in parallel, safely | brat | Agent Infra & Runtime | Rust |
| ship a local agent to production as an API | fastagentic | Agent Infra & Runtime | Python |
| run a task queue without Redis or RabbitMQ | fastworker | Agent Infra & Runtime | Python |
| automate workflows from one Go binary | m9m | Agent Infra & Runtime | Go |
| manage many coding-agent CLIs and their costs | ringlet | Agent Infra & Runtime | Rust |
| let an agent query a database, safely | ormai | Agent Infra & Runtime | Python |
| charge for MCP tools across payment rails | mcp-pay | Agent Infra & Runtime | Rust |
| add durable memory to an LLM chat app | memorg | Agent Infra & Runtime | Python |
| sandbox an agent’s files and shell | agentvfs | Agent Infra & Runtime | Rust |
| self-host a multi-channel agent gateway | openclawOS | Agent Infra & Runtime | TypeScript |
| share one agent gateway across tenants | openclawMU | Agent Infra & Runtime | TypeScript |
| embed a fast, sandboxed OpenClaw runtime | openclaw-rs | Agent Infra & Runtime | Rust |
| make your files an AI-searchable knowledge base | stratafs | Agent Infra & Runtime | TypeScript |
| make a Google ADK agent regulator-ready | regulus | Agent Infra & Runtime | Java |
| gate finance agents behind approval policy | closegate | Agent Infra & Runtime | Python |
| build AI agents inside a Flutter app | agentlib | Agent SDKs | Dart |
| run AI agents in the browser | linda | Agent SDKs | TypeScript |
| build agents with humans driving the state | onto | Agent SDKs | TypeScript |
| embed a tiny native agent via a C ABI | tokenworm | Agent SDKs | Zig |
| run a governed AI assistant on a bond desk | BondFoundry | Agent SDKs | Python |
| automate phone tasks with an on-device agent | ukkin | Agent SDKs | Dart |
| run your Jest suite far faster, unchanged | rjest | Dev & Build Tools | Rust |
| cut pytest startup and collection overhead | rpytest | Dev & Build Tools | Rust |
| make a huge git monorepo feel instant | gity | Dev & Build Tools | Rust |
| track issues in-repo with no server | grite | Dev & Build Tools | Rust |
| get AI-generated help at your shell Tab key | skillterm | Dev & Build Tools | Rust |
| cut warm Ninja/CMake build times | rninja | Dev & Build Tools | Rust |
| install Homebrew packages, much faster | stout | Dev & Build Tools | Rust |
| manage stacked pull requests on GitHub/GitLab | stkd | Dev & Build Tools | Rust |
| fix curl 403s without rewriting anything | recurl | Dev & Build Tools | Rust |
| download from sites that block wget | rewget | Dev & Build Tools | Rust |
No product matches that search.
How we build
The same principles run through all 36 products.
1
Drop-in by default
Keep the API, CLI or wire format you already use. One import or one binary swap — no rewrites, no config.
2
Rust where it matters
We profile, isolate the hot path, and rewrite it in Rust via PyO3 or a native binary. Native speed under the same surface.
3
Open source, no lock-in
Every product is MIT licensed and public. No accounts, no telemetry, no closed tiers — the CLI or library is the whole thing.
4
One sharp idea each
Small products that do one thing well and compose. Each lives at its own subdomain with its own docs and benchmarks.
Start with a single import
All 36 products are open source and MIT licensed. Every README, changelog and benchmark is in the open.
Frequently asked questions about Neul Labs
What is Neul Labs?
Neul Labs (Neul Labs Limited, UK Companies House #SC767862) is a UK-based open-source studio building high-performance infrastructure for the agentic era. It ships 36 MIT-licensed products across four families: Rust-accelerated drop-ins for Python AI frameworks, agent infrastructure and runtimes, agent SDKs, and developer/build tools. Each product lives at its own <name>.neullabs.com subdomain.
What does Neul Labs build?
36 open-source products. Rust accelerators (fast-litellm, fast-langgraph, fast-crewai, fast-axolotl, agentmatic); agent infra and runtimes (brat, fastagentic, fastworker, m9m, ringlet, ormai, mcp-pay, memorg, agentvfs, openclawOS, openclawMU, openclaw-rs, stratafs, regulus, closegate); agent SDKs (agentlib, linda, onto, tokenworm, BondFoundry, ukkin); and dev/build tools (rjest, rpytest, gity, grite, skillterm, rninja, stout, stkd, recurl, rewget). All MIT licensed.
Where can I find each product?
Every product lives at its own subdomain in the form <name>.neullabs.com — for example fast-litellm.neullabs.com or agentvfs.neullabs.com. Documentation is at docs.neullabs.com/<name>/ and source is at github.com/neul-labs/<name>.
Is Neul Labs open source?
Yes. All 36 products are MIT licensed and public at github.com/neul-labs. There are no closed tiers today.
How do I choose which Neul Labs product to use?
Start from the problem. To speed up an existing Python AI library use a Rust accelerator (fast-litellm, fast-langgraph, fast-crewai, fast-axolotl). To run or govern agents in production use the infra layer (fastagentic, agentvfs, ormai, memorg, mcp-pay). To build an agent use an SDK (agentlib for Flutter, linda for the browser, tokenworm for a native binary). To speed up your dev loop use rjest, rpytest, rninja or stout. The "Find a tool" decision table on the homepage maps each need to a product.
Is Neul Labs the same as Neuland Laboratories?
No. Neul Labs (Neul Labs Limited, UK Companies House #SC767862) is a UK open-source software studio. Neuland Laboratories Ltd. is a separate Indian pharmaceutical manufacturer (NSE: NEUL). The two companies have no connection.