01 Rust + Python
Rust Accelerators
fast-litellm
Drop-in Rust acceleration for LiteLLM. Faster connection pooling, rate limiting, and dramatically lower memory use.
Accountable engineering partner · Edinburgh, UK
Put AI to work— and make it dependable.
We help teams launch production AI, automate demanding workflows, and solve the integration, performance and control problems that hold delivery back.
Rust · Python · TypeScript
Source · tests · runbooks
Engagement / 01 From stuck to accepted
active
01
Recognise the situation Consequence · owner · deadline
qualify 02
Buy a bounded first step Fixed fee · written scope · exclusions
scope 03
Deliver against acceptance Tests · measurements · safe failure
build 04
Prove it and hand over Before and after · runbooks · ownership
accept Legal entity Neul Labs Limited
Companies House SC767862
Established 2023
Delivery UK & international
Which situation are you in?
Start from the problem, not the technology.
Most enquiries fit one of four situations. Each leads to a defined first step with a published price, and sometimes to the honest answer that an existing product or no project is the better choice.
01
Launch or rescue an AI system
A prototype that cannot launch, or a pilot that is losing confidence. We define readiness, measure the gap and close it without rebuilding what works.
Production AI and project rescue ↗ 02
Automate a business workflow
A recurring process that takes days, mostly waiting and re-keying. We connect the systems, validate early and measure the new turnaround.
Workflow automation ↗ 03
Fix performance, integrations or controls
Latency or cost blocking growth, a missing connector delaying a contract, or a security review your AI system cannot yet pass.
Compare specialist routes ↗ 04
Bring in senior engineering capacity
A hands-on AI engineer or architect inside your team for a defined term, with explicit ownership, priorities and a planned handover.
Embedded AI engineering ↗ Delivery outcomes
Measured results, honestly defined.
Anonymised, client-approved results from Neul Labs engagements. Each figure is elapsed turnaround, measured the same way before and after the change.
All case studies ↗ A UK employer with 250–1,000 staff
48 hours → 40 minutes
Payroll preparation
How automated input collection and validation cut payroll preparation turnaround from 48 hours to 40 minutes for a UK employer with 250–1,000 staff.
Read the case study ↗ An independent animation studio with under 50 staff20 days → 1 week
Production tracking and stage handoffs
Connecting production tracking and tools cut workflow turnaround from 20 days to one week for an independent animation studio with under 50 staff.
Read the case study ↗ How you buy
Buy a decision before a build.
Every engagement has a defined deliverable and a price you can see before the first call. Uncertain problems start with a review; well-defined ones are quoted directly as a sprint.
Compare engagements and pricing- 01
Readiness or feasibility review · £3,000–£6,000
Is this viable, and what is stopping it? Findings, a baseline and acceptance criteria in 1–2 weeks.
- 02
Focused engineering sprint · £15,000–£35,000
Can you fix or deliver this particular boundary? Working code, tests and measured results in 2–6 weeks.
- 03
Production delivery project · quoted after discovery
Can you own this defined system outcome? A deployed, accepted system, staffed only with capacity we can commit.
- 04
Ongoing engineering support · monthly, quoted
Can you maintain the system and help us improve it? Upgrades, evaluation runs and reserved change capacity, scoped to what we can genuinely cover.
Public engineering
Inspect how we build before you trust us.
Neul Labs publishes its own infrastructure in the open. It shows how we engineer. It is not a client case study, and you never have to adopt it to work with us.
View the complete catalogue ↗ 02 Rust
Agent Infra & Runtime
brat
Multi-agent harness for AI coding tools. Crash-safe state, parallel execution, one CLI.
03 Python
Agent Infra & Runtime
ormai
Give your AI agents database access without the risk. Policy-enforced ORM tools for Python and TypeScript.
04 Rust
Dev & Build Tools
stout
10–100× faster, Rust-based drop-in replacement for Homebrew.
The network
Every project has its own address.
This site is the hub. Each catalogue entry publishes its own
marketing site at <project>.neullabs.com, its
reference documentation under docs.neullabs.com, and
its source in the neul-labs GitHub organisation. The
table below is the full map, so a reader — or a retrieval engine —
can get from this page to any project in one hop.
Rust Accelerators
Drop-in Rust replacements that speed up the Python AI libraries you already use.
| Project | Language | Use it to | Project site |
|---|---|---|---|
| fast-litellm | Rust + Python | speed up LiteLLM without changing your code | fast-litellm.neullabs.com |
| fast-langgraph | Rust + Python | accelerate LangGraph checkpoints and state | fast-langgraph.neullabs.com |
| fast-crewai | Rust + Python | make an existing CrewAI crew faster | fast-crewai.neullabs.com |
| fast-axolotl | Rust + Python | fix Axolotl data-loading RAM/OOM at scale | fast-axolotl.neullabs.com |
| agentmatic | Rust + Python | run LangGraph-style agents on a Rust engine (source status pending) | agentmatic.neullabs.com |
Agent Infra & Runtime
The runtime layer agents need: orchestration, memory, sandboxes, data access, payments, governance.
| Project | Language | Use it to | Project site |
|---|---|---|---|
| brat | Rust | run several coding agents in parallel, safely | brat.neullabs.com |
| fastagentic | Python | ship a local agent to production as an API | fastagentic.neullabs.com |
| fastworker | Python | run a task queue without Redis or RabbitMQ | fastworker.neullabs.com |
| m9m | Go | automate workflows from one Go binary | m9m.neullabs.com |
| ringlet | Rust | manage many coding-agent CLIs and their costs | ringlet.neullabs.com |
| ormai | Python | let an agent query a database, safely | ormai.neullabs.com |
| mcp-pay | Rust | charge for MCP tools across payment rails | mcp-pay.neullabs.com |
| memorg | Python | add durable memory to an LLM chat app | memorg.neullabs.com |
| agentvfs | Rust | sandbox an agent’s files and shell | agentvfs.neullabs.com |
| openclawOS | TypeScript | self-host a multi-channel agent gateway | openclawos.neullabs.com |
| openclawMU | TypeScript | share one agent gateway across tenants | openclawmu.neullabs.com |
| openclaw-rs | Rust | embed a fast, sandboxed OpenClaw runtime | openclaw-rs.neullabs.com |
| stratafs | TypeScript | make your files an AI-searchable knowledge base | stratafs.neullabs.com |
| regulus | Java | make a Google ADK agent regulator-ready | regulus.neullabs.com |
| closegate | Python | gate finance agents behind approval policy | closegate.neullabs.com |
Agent SDKs
Frameworks for building agents — on the phone, in the browser, in a C ABI, or on a governed desk.
| Project | Language | Use it to | Project site |
|---|---|---|---|
| agentlib | Dart | build AI agents inside a Flutter app (source status pending) | agentlib.neullabs.com |
| linda | TypeScript | run AI agents in the browser (source status pending) | linda.neullabs.com |
| onto | TypeScript | build agents with humans driving the state (source status pending) | onto.neullabs.com |
| tokenworm | Zig | embed a tiny native agent via a C ABI (source status pending) | tokenworm.neullabs.com |
| BondFoundry | Python | run a governed AI assistant on a bond desk (source status pending) | bondfoundry.neullabs.com |
| ukkin | Dart | automate phone tasks with an on-device agent | ukkin.neullabs.com |
Dev & Build Tools
Faster tests, builds, git, packaging and networking for agent-era engineering.
| Project | Language | Use it to | Project site |
|---|---|---|---|
| rjest | Rust | run your Jest suite far faster, unchanged | rjest.neullabs.com |
| rpytest | Rust | cut pytest startup and collection overhead | rpytest.neullabs.com |
| gity | Rust | make a huge git monorepo feel instant | gity.neullabs.com |
| grite | Rust | track issues in-repo with no server | grite.neullabs.com |
| skillterm | Rust | get AI-generated help at your shell Tab key (source status pending) | skillterm.neullabs.com |
| rninja | Rust | cut warm Ninja/CMake build times | rninja.neullabs.com |
| stout | Rust | install Homebrew packages, much faster | stout.neullabs.com |
| stkd | Rust | manage stacked pull requests on GitHub/GitLab | stkd.neullabs.com |
| recurl | Rust | fix curl 403s without rewriting anything | recurl.neullabs.com |
| rewget | Rust | download from sites that block wget | rewget.neullabs.com |
Source status is audited separately from the catalogue. Seven entries — agentmatic, agentlib, linda, onto, tokenworm, BondFoundry, skillterm — had a configured source URL that did not resolve at the last audit, so their repository link is withheld until it is restored or the entry is revised. Their project sites and catalogue records remain published, and every benchmark claim on this network is workload- and version-specific: the linked method is the authority, not the headline number.
Delivery partners
You have the customer. We take the hard work package.
For consultancies and platform vendors with a named customer engagement: integrations, production hardening, evaluation and performance work, delivered with explicit acceptance and handover.
NEUL / LABS Engineering partner
01 Consultancies Specialist work packages · white-label
02 Platform vendors Connectors · MCP · SDKs · enablement
03 R&D programmes Prototypes · evidence · transfer
Have a difficult system?
Bring us the boundary that keeps breaking.
Tell us what has to work, what is failing today and what evidence would make the next decision easier.
Start with the problemWhat does Neul Labs do?
Neul Labs helps teams put AI into production, automate demanding workflows, and fix the performance, integration and control problems that block adoption. Work is sold as bounded engagements — fixed-fee reviews, fixed-scope sprints, delivery projects, embedded engineering and ongoing support.
Can Neul Labs improve an existing AI system without rebuilding it?
Yes. An engagement begins with the existing workflow, stack and failure cases. The first step is to establish a baseline and identify the limiting components; the implementation then targets those components, with agreed tests, rollback and handover requirements.
How much does an engagement cost?
Indicatively, a readiness or feasibility review is £3,000–£6,000 and a focused engineering sprint is £15,000–£35,000, excluding VAT where applicable. Larger delivery projects are quoted after discovery. The fee for each engagement is fixed in writing.
Do we have to use Neul Labs open-source projects?
No. The open-source work is evidence of how we engineer and a set of options to assess on their merits. Many engagements use none of it, and the best answer is sometimes an existing platform, a small integration or no project at all.
Where is Neul Labs based?
Neul Labs Limited is registered in Scotland (SC767862) and works remotely with organisations in the United Kingdom and internationally.
Is Neul Labs the same as Neuland Laboratories?
No. Neul Labs Limited is a UK software engineering company. Neuland Laboratories Ltd. is a separate Indian pharmaceutical manufacturer. The two companies have no connection.