Thing Event System

The governance layer for agentic commerce

AI agents are making purchases, managing subscriptions, and moving money — with no audit trail. Verifiable Intent proves the transaction was authorised. TES proves the agent behaved. An event-sourced ledger that gives enterprises, agents, and regulators a shared, auditable record of every entity they touch.

Try the Playground
tes events --stream

Building on standards from

MastercardGoogleCloudflare

The Problem

AI agents are making purchases, managing subscriptions, and moving money — with no audit trail. The EU AI Act (August 2026) demands governance. TES provides it.

How it works

Three primitives, infinite workflows

Every entity lifecycle follows the same pattern: emit events, project state, settle value.

01

Emit

Correlation IDs link the full decision journey across sessions, agents, and entity types.

await tes.emit("agent_session.action_taken", {
  session_id: "ses_7f2a",
  action: "select",
  entity_id: "prod_b2",
  agent_id: "agent_007",
}, {
  correlation_id: "cor_x7y8z9",
  idempotency_key: "idem_abc123",
});
02

Project

Query any entity — agent sessions, invoices, subscriptions, items — and get the full, auditable truth.

const session = await tes.project("agent_session", "ses_7f2a");

// Returns derived state:
// {
//   status: "completed",
//   actions: ["browse", "compare", "select"],
//   vi_credential: "vi_m4c9",
//   compliant: true,
//   outcome: "purchase_completed"
// }
03

Settle

What the agent did, why it was allowed, and what happened after.

// Settlement triggered after VI verification
// TES records the complete lifecycle

{
  type: "settlement.completed",
  correlation_id: "cor_x7y8z9",
  data: {
    protocol: "agent_pay",
    vi_credential: "vi_m4c9",
    net_amount: 242.49,
    fee: 7.50,
    ledger_entries: [
      { debit: "usr_jane", credit: "merchant_acme", amount: 249.99 }
    ]
  }
}

Use Cases

One governance layer, every entity lifecycle

Agent sessions, invoices, subscriptions, metered APIs, physical goods — if it has a lifecycle, TES governs it.

Full-journey observability for AI agents

Every agent action — browse, compare, select, pay — captured as immutable events. Verifiable Intent (Mastercard + Google) proves authorisation. TES proves behaviour. EU AI Act compliant by design.

startedaction_takenvi_credential_attachedpayment_initiatedcompleted
vi_credential_attachedcompleted

Architecture

The protocol stack

TES sits between applications and settlement — capturing every agent action, entity state change, and compliance event as an immutable record.

TES — Thing Event System

The governance layer that records every agent action as an immutable event. Projections derive current state. Correlation IDs link related events across agents and sessions. Compliance queries answer: what did this agent do, and was it authorised?

// TES stores immutable events and derives state
{
  id: "evt_a1b2c3",
  type: "agent_session.action_taken",
  timestamp: "2026-02-17T10:00:00Z",
  correlation_id: "cor_x7y8z9",
  subject: { type: "agent_session", id: "ses_7f2a" },
  data: { action: "compare", vi_credential: "vi_m4c9" }
}

// Query: what did this agent do?
const trail = await tes.query("agent_session", "ses_7f2a");
// → [action_taken, action_taken, vi_credential_attached, ...]

Pricing

Start building with TES

Pay only for what you use. No minimum commitments.

Developer

Free

Get started with TES. Perfect for prototyping and development.

  • 10,000 events / month
  • 5 entity types
  • 7-day event retention
  • Community support
  • Playground access

Growth

$0.001/ event

Production-ready event processing with full retention and projections.

  • Unlimited events
  • Unlimited entity types
  • 90-day event retention
  • Projections API
  • Agent Pay, x402 + Stripe settlement
  • Email support

Enterprise

Custom

Dedicated infrastructure, SLAs, and white-glove onboarding.

  • Everything in Growth
  • Unlimited retention
  • Dedicated infrastructure
  • Custom SLAs (99.99%)
  • SSO / SAML
  • EU AI Act compliance reports
  • Dedicated support engineer
Contact sales

Denial

$40M+/ incident*

The classic approach. No transparency, no visibility, unlimited liability.

  • Pay nothing until August 2026
  • Up to $40M in regulatory fines
  • Up to 7% of global annual revenue per breach
  • Mandatory incident disclosure to supervisory authority
  • Full regulatory audit of all agent activity
  • No audit trail to defend your decisions

* EU AI Act enforcement begins August 2026. Maximum penalty of €35M (approx. $40M) or 7% of global annual revenue per enforceable incident — whichever is higher. Colorado AI Act effective February 2026.

Get started

Ready to govern your AI agents?

Start building with TES in minutes. Free tier includes 10,000 events per month.

Read the docs