Solon · The Product

Policy.
As code.

Every payment intent your agents make is evaluated against your policies — before it's signed. Real code, real evaluation, leaving a complete audit trail.

— What we don't do

We don't touch
your keys.

Private keys stay with your wallet, your MPC setup, or your custodian. Solon never holds a complete key. This is a structural guarantee, not a policy commitment.

— What we do do

We decide what
your agents spend.

Policy engine, approval workflow, compliance screening, audit trail, multi-agent coordination, ERP integration. Your enterprise rulebook, translated into machine-executable code.

01 · The Engine, Live

Every transaction,
examined in code.

Four real transactions, looping. Watch each one move through the policy engine in real time.

solon@control-plane ~/policies · evaluate.ts live
UTF-8 · TypeScript · policy-dsl v2 engine · --:--:-- · evt 0
02 · How it works

Three steps.
Microseconds each.

Step 01

Intent arrives.

Your agent emits a payment intent — via x402, A2A, or any supported protocol. Solon's adapter parses the request into a structured object: amount, recipient, chain, context.

Step 02

Policies evaluate.

Each rule runs against the intent: spend limits, whitelist matching, risk scoring, compliance checks, ERP mapping. Every decision is recorded — agent ID, prompt hash, model version, timestamp.

Step 03

Signed, held, or stopped.

Approve → forward to your wallet for signing and settlement. Hold → queue for human approval. Reject → log and block. Either way, the audit trail is complete and immutable.

Built into a five-layer stack.

Solon is the third layer. See how it fits with protocols, infrastructure, and on-chain settlement.