Cajal

Scaling formal verification to accelerate scientific discovery

Winter 2026ActiveB2BB2BBig DataAISan Francisco, CA, USA

Company

Cajal (YC W26) is massively scaling formal verification to accelerate scientific discovery. We deploy superhuman AI mathematicians to high-impact applied domains, starting with quantum computing and finance. We do this with Lean - a framework that allows us to formally verify any mathematical statement, grounding AI in truth and validating the tools discovered by our systems.

Verdict

Medium Signal

Market Opportunity

Formal verification for quantum computing and finance is a real and potentially large B2B market, but the ICP is extremely narrow today and enterprise sales cycles in both verticals are long. The 'scientific discovery' framing is broad but the stated starting verticals (quantum + finance) are credible high-value domains willing to pay for correctness guarantees. TAM is real but likely sub-$1B for formal verification tooling specifically.

Medium Signal

Founder Signal

Luke Johnston has an MPhil from Cambridge in ML/Computational Neuroscience (First Class) and a UCL research internship at the Gatsby Unit — solid academic ML background but limited industry shipping experience; most recent roles are research intern and part-time Google course instructor. Pedro Nobre co-founded Vertebra (AI consulting, 2022-2024) and has some consulting/engineering background, but no deep formal verification or Lean expertise is evident. Both came through Entrepreneurs First before YC W26, suggesting the idea is relatively nascent.

Medium Signal

Competition

No competitor data was returned, but the space includes established players like Galois, Certora (smart contract verification), and academic Lean/Coq tooling ecosystems. Microsoft Research has deep Lean investments. The AI-augmented formal verification angle (e.g., Google DeepMind's AlphaProof) is increasingly crowded with well-resourced players. Differentiation claim is plausible but unproven.

Low Signal

Product

Website is completely broken with a client-side application error — zero product visible. No demo, no API docs, no pricing, no customer logos, no testimonials, and no usage metrics anywhere in the data. Pure vaporware at this stage.

OverallC Tier

Cajal is an early-stage bet on a technically interesting but extremely hard problem — scaling formal verification with AI — with a team whose credentials lean academic rather than deeply industry-tested. The website is broken, there's zero evidence of customers, revenue, or even a working demo, and no press coverage exists. Luke's ML/neuroscience background at Cambridge and UCL is solid but formal verification (Lean specifically) is a highly specialized domain neither founder has demonstrated deep expertise in publicly. Pedro's prior startup (Vertebra) shows some operator experience but was an AI consulting shop, not a deep-tech product company. The idea is genuinely differentiated if they can execute, but right now this looks like a research project searching for product-market fit.

Active Founders

Luke Johnston

Founder

Working on formal verification and AI. Background in machine learning and neuroscience at research labs in oxford/cambridge/ucl. Learn more at https://caj.al/ and feel free to get in touch at luke@caj.al

Pedro Nobre

Founder

Working on formal verification and AI.

Cajal

TierC Tier

BatchWinter 2026

Team Size2

StatusActive

LocationSan Francisco, CA, USA