AI is getting faster at finding vulnerabilities. Does it catch what auditors miss, or replace them? Most security tools look for what's broken. Formal methods prove what's correct. Why aren't more protocols using them? L2s promise composability. You still wait for finality. What if you didn't have to? Ethereum's next upgrade ships in months. Is the DA layer built for a post-quantum world? What needs to change in the execution layer? AI agents will need to make payments. How do you keep them private without breaking compliance? Seven talks. Rekt Summit and EthCC.(Nethermind)