Bulgaria’s first publicly proven open agentic-commerce system passes an external production test

Bulgaria’s first publicly proven open agentic-commerce system passes an external production test — Brand Design Ltd.

On 21 September 2026, an external AI agent discovered 101Ts3t without a merchant hint, completed a real payment and received a public PASS.

As of 21 September 2026, this is Bulgaria’s first publicly proven operational open agentic-commerce system. The result is publicly verifiable, and the proof package is attested by GitHub/Sigstore.

One continuous agentic-commerce path

A third-party participant gave its AI agent the exact neutral task issued by the public test. The prompt contained the service code 101Ts3t and a maximum budget of EUR 1.00. It did not contain a merchant name, domain, destination or checkout link.

  1. Discovery: the agent found the public offer independently.
  2. Selection: it selected 101Ts3t at the published price of EUR 0.99.
  3. Order and payment: it placed the order and completed a real production payment through its authorised wallet.
  4. Fulfilment and verification: the merchant completed fulfilment and the verifier finalised all four protocol gates as PASS.

Why the result is externally verifiable

The test rules were published before the transaction. The public verification record reports a finalised PASS; the privacy-safe proof belongs to the public test package. The GitHub/Sigstore attestation anchors the package files, while the Zenodo implementation record, version 1.3 preserves the implementation. All sources refer to test qst_44e153fc16fe3fa91520e5f6922ae4c9.

What the successful test demonstrates

The production run demonstrates the complete path from discovery and trust to selection, order, payment, fulfilment and public verification. The buyer’s agent was free to discover the merchant through public information and was not required to use MCP.

Read the practical explanation of agentic commerce, see the Agentic Commerce service, or inspect the public catalogue for AI agents.