Researcher-led security reviews driven by large-scale parallel agent analysis of code

Grego AI sells security review as an engagement, not a scanner subscription. A senior researcher maps the scope with the customer's engineering team, capturing the invariants the system is supposed to preserve and the trust boundaries it assumes, because in protocol code the interesting bugs are violations of intent rather than pattern matches. The platform then runs agents in parallel across the code paths, each pursuing an exploitation hypothesis, an approach the company calls Deep Invariant Analysis and aims at flaws that emerge from several layers of dependencies interacting. Reviews follow those dependencies past the stated boundary when behaviour depends on external code, which is where a nominally in-scope contract usually breaks.

The deliverable is a report with findings ranked by severity, each accompanied by a reproducible attack sequence, an estimate of economic impact where it can be calculated, and remediation guidance, followed by targeted retesting. The platform handles Solidity, Rust, Go, TypeScript, Move, and CosmWasm, and customers use it before, during, or after a traditional audit as an independent second opinion. Grego AI was founded in 2024 by CEO Justus Hanna, a highly ranked bug bounty researcher who works publicly as riptide, and CTO Gregorio Maspero, with a seed round led by cyber Fund and backing that includes Vercel CEO Guillermo Rauch. The company says it holds the top position for an AI security tool on the Immunefi and Hackenproof leaderboards, and that one of its findings averted a loss of $27.7M in a blockchain protocol.