Why does my AI-written smart contract ship with no properties for a verifier to check?
Opportunity
Formal verification of smart contracts requires invariants and pre/post conditions describing what the code is supposed to do. AI agents write Solidity fluently but produce no property specifications alongside the code; auditors still write those specs by hand, when they write them at all. PropertyGPT, which won the Distinguished Paper Award at NDSS 2025, was the most serious attempt to generate specs automatically from code using LLMs. It reached only 80% recall against expert-written properties, leaving critical invariants missing and contracts going to mainnet without a complete specification. No production development toolchain generates property specifications as a standard step, which means formal verifiers like the Certora Prover are rarely applied even when the code itself is AI-generated.
Why it matters
Automatically generating correct, complete property specs alongside AI-written contracts is what finally closes the loop between AI-assisted coding and machine-verified contract correctness.
How I score the opportunity
The Opportunity Score is my own read, not a measurement: how much it hurts, how often it bites, and how little exists to solve it today. Higher means I think it is more worth building.
How much pain it causes when it shows up.
How often people actually run into it.
How little good tooling exists for it today.
More problems worth solving
What does an AI agent's bank account actually look like?
AI x CryptoCan an on-chain organization run by agents avoid becoming a scam machine?
AI x CryptoHow do you prove a photo or a voice is real without a platform vouching for it?
AI x CryptoWhy is on-chain identity either nothing or your entire life?
AI x CryptoHow do I audit which agent acted under my identity across a delegation chain?
AI x CryptoWhy does running my prompt inside an encrypted VM not actually keep it private?