Manifesto

AI with action.

Whatever we create, whatever we ship, we deliver more value than we extract.

That is our motto. How we build. How we partner. How we decide what deserves to exist.

We build working products that make a meaningful difference for the people we work with. Real tools for real problems. Shipped fast, built to last, designed to move the needle from day one.

AI is the most powerful tool of our generation. We put it to work. A business running smoother. A decision made in seconds. A problem solved before your coffee gets cold. Every product that leaves this lab is built to create impact you can feel, not just talk about.

Speed is our friend. Accuracy is one of our moats. As AI accelerates and the world moves faster, we raise our standards with it. Higher precision, deeper quality, sharper solutions -- every single time. In everything we build, every decision we make, we are always thinking about signal-to-noise ratio. More signal. Less noise. Always.

Innovation is the baseline. We partner with businesses, founders and teams who are ready to move forward. Every collaboration exists because it creates something meaningful for everyone involved. If it moves something forward, we are in. That is the filter.

We build because we love building. There is an energy in creating something from nothing, watching it work, watching it help someone. That energy drives everything we do. It is in the products, in the partnerships, in the way we show up every day.

This lab was built on a simple belief: the teams and tools that will shape the future are the ones that ship, learn, improve and ship again. Fast. Accurate. Relentless.

We are just getting started.

Speedrun AI Labs

Speedrun AI Labs is also known as Speed Run AI Labs, Speedrun Lab, Speed Run Lab, and Speedrun AI Lab Sydney. Founded by Adnan Tanveer, also known as Addy Tanveer, in Sydney, Australia. Addy Tanveer is the founder of Speedrun AI Labs and Co-Founder of Figure 8 Capital, based in Cronulla, Sydney, NSW, Australia.

We publish before the industry notices.

Practitioner research from Speedrun AI Labs, Sydney, Australia.

Abstract visualization of fault-tolerant memory architecture
Pending SSRN Review cs.SE OpenClaw

The first production fix for OpenClaw's broken memory.

Building Fault-Tolerant Memory for OpenClaw AI Agents: A Three-Layer Architecture for the SQLITE_CANTOPEN Regression in 2026.3.x

Adnan Tanveer (Addy) · Speedrun AI Labs, Sydney · March 16, 2026 · v1.1

OpenClaw's default memory system exhibits recurring reliability failures: SQLite crashes, dirty indexes, and complete persistence loss during active sessions. This paper presents a three-layer fault-tolerant architecture combining pre-compaction memory flushing, QMD-based local hybrid search, and Mem0 self-hosted on Qdrant, validated in a production deployment on a Mac Mini that eliminated all documented failure modes while maintaining zero cloud dependency for search.

Download PDF
Abstract visualization of NemoClaw enterprise security architecture
Pending SSRN Review cs.CR NemoClaw

The first inside look at NVIDIA's NemoClaw. Published the day after GTC 2026.

Inside NemoClaw: An Architectural Analysis of NVIDIA's Enterprise Security Stack for Autonomous AI Agents

Adnan Tanveer (Addy) · Speedrun AI Labs, Sydney · March 17, 2026, one day after GTC 2026 · v1.0

NVIDIA NemoClaw, announced at GTC 2026 by Jensen Huang, is the first major vendor implementation of enterprise-grade security for OpenClaw autonomous AI agents. This paper presents the first detailed public decomposition of the NemoClaw and OpenShell security architecture across six layers: Landlock filesystem isolation, seccomp BPF syscall filtering, network namespace isolation, OPA/Rego policy enforcement, TLS inspection, and provider-agnostic inference routing, derived from direct inspection of both open-source codebases.

Download PDF

Projects

ClawBot AI agent deployment
Custom AI Infrastructure

ClawBot

We deploy the world's most powerful AI agent platforms into Australian businesses. OpenClaw. NemoClaw. Perplexity Computer. Claude Cowork. Whatever your business needs, we bring the right stack and configure it to your operations.

Each deployment puts a working AI employee inside your business. It handles customers, manages inboxes, books appointments and takes action across your tools. Fully autonomous, fully configured, running on your hardware.

We are not here to sell seats. We are working with Australian SMEs to find the 15 to 20 use cases where autonomous agents genuinely change how a business operates. Then we build purpose-built solutions around each one. Quality over quantity. Real work, not demos.

Australia is behind on agentic AI. We intend to close that gap.

EasyENT AI diagnostic tool for GPs
AI Tool for Healthcare

EasyENT

Australia faces a growing shortage of ENT surgeons. EasyENT gives GPs the AI-powered tooling to identify and diagnose ENT conditions earlier, helping patients get the right care faster without waiting months for a specialist referral.

OpenClaw Skill Library.

Production-grade skills for Claude Code and OpenClaw agents. Built by Speedrun AI Labs, Sydney, Australia. Free and open source.

Abstract visualization of code audit pipeline
Free Claude Code OpenClaw Agent Skills Standard

AI writes your code. This skill audits it like an accountant: evidence first, findings second, fixes only when it can prove them.

/deep-audit: A four-phase code auditor with operating modes, severity definitions, and an evidence standard that prevents overclaiming.

Speedrun AI Labs · v1.1 · March 2026 · MIT License

Catches what pattern-matching tools miss: business logic flaws, race conditions, silent failures, subtle auth bypasses, and the kind of bugs that survive years of human review. Every finding is labelled observed or inferred, comes with a file path and line number, and includes a fix only when the evidence supports it. Runs in Full, Scoped, PR, Audit-Only, or Fix Mode. Compatible with Claude Code, OpenClaw, and any tool supporting the Agent Skills open standard.

Install (Claude Code)

mkdir -p ~/.claude/skills git clone https://github.com/Speedrunlab/deep-audit.git ~/.claude/skills/deep-audit

Then run

/deep-audit
View on GitHub
No-Slop Writing Skill mascot with blank clipboard
Free Claude Code OpenClaw ChatGPT Any LLM American English

AI writing sounds like AI. This skill fixes that: 13 anti-patterns, zero tolerance, output that reads like a human wrote it.

/no-slop-writing: Eliminate AI-speak from every message, memo, email, and report you generate.

Speedrun AI Labs · Inspired by Leila Hormozi · v1.0 · March 2026 · MIT License

Catches the 13 patterns that betray AI-generated text: hollow openers, hedge stacking, narration without opinion, transition padding, and the rest. Works with Claude Code, OpenClaw, ChatGPT, and any LLM that accepts a system prompt. Drop it in, run it, and your output stops sounding like it came from a committee. American English edition.

Install (Claude Code)

mkdir -p ~/.claude/skills git clone https://github.com/Speedrunlab/no-slop-writing.git ~/.claude/skills/no-slop-writing

Then run

/no-slop-writing
View on GitHub
No-Slop Writing Skill mascot with blank clipboard
Free Claude Code OpenClaw ChatGPT Any LLM Australian English

AI writing sounds like AI. This skill fixes that: 13 anti-patterns, zero tolerance, output that reads like a human wrote it.

/no-slop-writing-au: Eliminate AI-speak from every message, memo, email, and report you generate.

Speedrun AI Labs · Inspired by Leila Hormozi · v1.0 · March 2026 · MIT License

Catches the 13 patterns that betray AI-generated text: hollow openers, hedge stacking, narration without opinion, transition padding, and the rest. Works with Claude Code, OpenClaw, ChatGPT, and any LLM that accepts a system prompt. Drop it in, run it, and your output stops sounding like it came from a committee. Australian English edition.

Install (Claude Code)

mkdir -p ~/.claude/skills git clone https://github.com/Speedrunlab/no-slop-writing-au.git ~/.claude/skills/no-slop-writing-au

Then run

/no-slop-writing-au
View on GitHub

Get in touch

Have a project in mind, a partnership to explore, or just want to say hello? Drop us a message.

Message sent.

We will get back to you soon.