Testmachine
AI-driven Web3 security testing and token risk scoring platform
Performances
RD Popularity Index
RD Growth Index
Comparison
Follow Updates
Follow Lists
People
About Testmachine
TestMachine runs smart contract and on-chain security testing from an attacker's viewpoint. Its core component, Azimuth, is a reinforcement-learning agent that repeatedly tries combinations of function calls inside a high-fidelity EVM simulation of forked mainnet, and reports only vulnerabilities it has actually exploited and can demonstrate with working exploit code, which removes the bulk of false positives typical of static scanning. The same capability is packaged as Predator, a continuous monitoring engine for tokens that tracks behaviour appearing only after deployment, such as proxy upgrades, hidden privileges, blacklists and confiscation clauses.
On public benchmarks, Azimuth detects 86.3% of vulnerabilities in EVMBench at an observed false positive rate of about 2.7%, at roughly USD 100 of average cost per repository; the engine reproduced the full exploit path behind the roughly USD 196 million Euler Finance loss of March 2023, using two simulated accounts to bypass the separation between borrowing and self-liquidation. On the engineering side, its token scanning is integrated into Coinbase's centralised and decentralised trading pipelines, and it has partnerships with the Avalanche Foundation and Spectral Labs.
TestMachine disclosed a $6.5 million round in January 2026 led by BlockChange Ventures, New Form Capital, Decasonic and Delphi Digital, with Baboon, UDHC, Auros Global, Generative Ventures, Contango Digital and Santiago Santos participating. Its AI security engine Predator is integrated into Coinbase's risk process across centralised and decentralised exchange trading, monitoring more than one million tokens in real time; the company says it caught all $12 million of rug pulls in an 11,000-token sample.
Testmachine also published EVMBench results: an 86.3% detection rate, roughly 2.7% observed false positives and about $100 average cost per repository, with Azimuth reproducing exploitable bugs on forked state.
Future strategy and roadmap: Next, TestMachine will extend Azimuth from smart contracts to autonomous systems and code, launch the continuous-validation Agentic Security product, refresh its benchmark results quarterly, and plug exploit validation into listing risk checks at exchanges and issuers.
Andrew Kilbride — Co-founder & CEO; leads strategy and commercialization and drove Azimuth's integration into Coinbase's token security process.
Matthew J. Lewis — Co-founder & CTO; PhD in physics from the University of Michigan; former machine-learning role at General Dynamics and principal investigator on a DARPA program.
John Calabrese — Head of Product.
Elaine Murtagh — Head of Business Development.