Speak at The Fifth Elephant 2026 Annual Conference

To confidently deploy in production, you need a robust verification mechanism. Some verification mechanisms are easy. Wrong code doesn’t compile or pass good test cases. Wrong analysis doesn’t meet a post-condition - say a value range or a known aggregate. Wrong proofs don’t validate on LEAN.

But how far can we take it? When we have an insurance claim, could InsurLE help us verify its validity? Could Catala help us verify the legality of a contract?When an engineer generates a circuit, could we simulate it on SPICE? Use SHACL validated knowledge graphs to verify drug details?

In this workshop, you’ll explore a variety of different problems that are more verifiable than we thought. That lets us run build and deploy agents far more reliably. You will also be creating verification harnesses where none existed before - which is the bottleneck since agents have driven the generation cost to almost zero.

This session is for AI developers who need to ship verifiale agents into production - e.g. in a regulated or high-trust environment (finance, healthcare, government).

{Add the link to draft slides - PDF/PPT - with comments access - TODO: at least a TOC or blog series}

{Add the link to 2-min elevator pitch video - Um... skip, please?}