By bringing the computation model directly to the source database, AEM prevents data leakage while enabling complex, multi-party analytical insights.
Data providers install the lightweight AEM Node Agent on their local servers. You define the table schemas, custom pricing structures, and query rules. Only descriptive metadata is indexed on the central registry—your raw records remain safely at rest on your infrastructure.
Researchers, model builders, or business analysts explore AEM schemas. They write queries or upload Dockerized machine learning scripts (e.g., Python, PyTorch). The AEM gateway verifies access authorization, payment deposits, and checks compliance restrictions.
The approved compute task is dispatched directly to the data provider's Node. It executes within a hardware-secured Trusted Execution Environment (TEE) or sandboxed container. The code operates locally on the database, generates statistical parameters, and deletes the temp space.
The query outputs are packed, encrypted, and sent back to the consumer. Simultaneously, a cryptographic verification proof is committed to the AEM ledger. Smart contracts trigger automated payments, depositing the correct royalty fees directly into the data owner's account.
AEM Node agent integrations are regularly audited by leading cybersecurity firms. Code execution sandboxes prevent system calls, keeping host filesystems 100% read-only.
Whether you have data to publish, models to ship, or questions to ask we'd love to hear from you.
