Agent-to-Agent Demo
Any AI agent with TEE-backed identity can request data attestation from the Prova Trust Engine.
Agent Identity
Each agent gets a DID rooted in its TEE attestation. Identity = code hash + hardware quote.
dstack + W3C DIDData Provenance
MPC-TLS proves data came from the real source. Notary co-signs without seeing plaintext.
Props + TLSNotaryAttested Computation
AI model weights are pinned to the attestation. Exact model that ran is cryptographically bound.
Props Model HashAgent Identifies Itself
The requesting agent presents its TEE attestation (enclave hash + TDX quote) to the Prova Trust Engine.
Submit Data for Attestation
Agent sends raw data (e.g., clinical notes, financial records, legal documents) to the /v1/process endpoint.
TEE Processing
The Trust Engine processes data inside the enclave: AI formatting, deterministic validation, anonymization.
Attestation Bundle Returned
Agent receives the attested output with full cryptographic proof chain: TDX quote, model hash, MPC-TLS proof, credential destruction.
Downstream Verification
Any relying party can independently verify the attestation bundle at /verify — no trust in the requesting agent required.
Multi-Vertical Use Cases
Healthcare
EHR data → FHIR R4 Claims with HIPAA-compliant anonymization
Finance
Bank statements → SEC-compliant risk reports with PII stripping
Legal
Court records → Structured legal documents with privilege protection
Government
Agency data → Standardized forms with citizen PII protection
Insurance
Claims databases → ACORD XML with cross-regulation compliance
Supply Chain
ERP/logistics → Authenticated bills of lading and customs declarations