Research

Evidence for the systems agents depend on.

We study the boundaries that make financial agents useful: identity, wallet context, tool use, simulation, authorization, evaluation, and verifiable execution.

Identity

Keep account, wallet ownership, provider sessions, and delegated authority as separate proofs.

Evaluation

Measure model performance on executable onchain tasks, not only plausible text.

Execution

Study construction, simulation, policy, signing, settlement, and evidence as one system.

Published work

Research notes and benchmarks

01research · September 18, 2026Enabling Agentic Finance: Three Gates from Intent to Settlement

How the runtime, wallet, smart contract and block builder can constrain agent-built transactions.

02research · August 13, 2026The State of Execution Harnesses for Agentic Payments

Breaking down the seven-layer agentic payments stack and identifying the missing execution harness that makes financial agents efficient, reliable, and operational.

03engineering · July 13, 2026One User, Many Wallets: Authentication Across the Onchain–Offchain Boundary

A design study in hybrid crypto authentication: why counting users is hard onchain, the four proofs to keep separate, and the patterns — and stack — that hold them apart.

04benchmark · June 4, 2026AomiBench: Benchmarking Frontier Models on Onchain Execution

A wallet-aware harness for measuring frontier agents on real onchain tasks, simulations, and chain-state evidence.