This topic covers emerging standards and protocols (exemplified by proposals such as ERC-8183) that enable “trustless” commerce between autonomous AI agents and human users or services. These standards specify on‑chain primitives (identity and DID binding, capability tokens/permissions, verifiable receipts, escrow/settlement hooks and off‑chain metadata references) so agents can negotiate, prove authorization, and settle payments or service agreements without bespoke integrations. Relevance in 2026: as agent deployments and enterprise assistant integrations proliferate, organizations need composable ways to audit intent, delegate limited capabilities, and reconcile billing across systems. Model Context Protocol (MCP) servers — the integration layer used in chat API integrations — are central to that stack. Example MCP implementations include Google Workspace and Microsoft 365 MCP servers (multi‑user OAuth 2.1/MSAL, Graph API), Salesforce MCP (OAuth, dynamic schema discovery), and Slack and Microsoft Teams MCP servers (stdio/SSE/HTTP transports, scoped modes). These servers provide the consent, context and audit trails that on‑chain standards rely on for trustworthy mappings between off‑chain activities and on‑chain settlements. Practical implications and trends: expect hybrid flows where MCPs supply authenticated action requests, verifiable credentials and event logs, while ERC‑style contracts handle escrow, settlement and public auditability. Key concerns are least‑privilege authorization, replay‑resistant attestations, privacy for off‑chain data and standard message schemas. For chat API integrations, adopting common capability tokens, DID bindings and receipt formats reduces bespoke engineering and enables interoperable, auditable commerce between agents and users.