Built by veterans from Lido, Nomura, P2P.org

Real-time underwriting engine for DeFi protocols

Given a protocol and an intended interaction, return a current underwriting assessment with confidence, reasoning, and what changed since last check. Continuously re-derived as governance, dependencies, and on-chain state shift.

Underwriting broke

Reports go stale in days

Governance votes, upgrades, and dependency shifts happen constantly. Human reports don't.

Static docs aren't a runtime signal

Machine consumers need structured, current risk state before signing, not a static document.

Volume crossed the threshold

100M+ agent-initiated transactions on Base in the last 12 months, up from near zero.

Consulting doesn't scale

Human-services businesses cover the top ten. DeFi is thousands of protocols, composed.

The result: risk answers that lag the assets they're supposed to protect.

The Product

One engine, three interfaces

An API. Given a protocol and an intended interaction, return a current underwriting assessment. The engine continuously re-derives as on-chain state, governance, and dependencies shift.

Continuously re-derived. Machine-readable. Composable-aware.

  • Deep assessments replacing manual reports
  • Pre-transaction checks called from any signing path

Protocol

Aave v3

Interaction

Supply USDC

Unified Engine

Deep Assessment

Human risk teams

Monitoring Stream

Live delta feed

Pre-tx Check

Agent signing path

How it works

01

Query

Protocol and intended interaction sent to the API from a signing path or risk workflow.

02

Assess

Engine re-derives from live on-chain state, governance events, and the dependency graph.

03

Return

Structured response with confidence, reasoning, and the delta since last check.

04

Monitor

Ongoing stream flags material changes to your existing exposure as they happen.

Built for the machine underwriting era

Continuous re-derivation

Not a PDF frozen in time. The engine re-runs as on-chain state, governance events, and dependencies shift, so the answer you get is the answer as of now.

Machine-native output

Structured assessments with confidence and reasoning, callable from any framework's signing path.

Coverage that compounds

Each new protocol integration improves assessments of everything connected to it via composability.

Calibrated against realized loss

Ground-truth exploit history, dependency graph, and governance timeline, so confidence scores are grounded, not guessed.

Continuous
Composable-aware
Calibrated
API-first

Who uses Unified

Agent frameworks

Solver networks & intent architectures

Pre-transaction risk check wired into the signing path, the default risk feed for agent-initiated flows.

Protocol teams

Live dependency monitoring

Track material changes across your integrations, upgrades, and dependencies as they happen.

Human risk teams

Treasuries, funds, DAOs

Deep assessments replacing manual reports. Lido today, expanding to funds and treasury committees.

Request Whitelist Access

We're onboarding design partners across three lanes: human risk teams, agent frameworks, and protocol teams. Tell us where you fit.