Private AI without cloud APIs: what must be isolated
Build private AI without external APIs by controlling inference, embeddings, retrieval, telemetry, updates and logs.
Removing the generation API is insufficient if embeddings, observability or updates still leave the declared perimeter.
Secondary dependencies are often invisible in prototypes and appear only during deployment or audit.
An outbound-flow inventory and negative network test should support every no-cloud-API claim.
What the architecture must make explicit
Privacy and sovereignty are properties of the executed system: network flows, data, telemetry, dependencies, updates, backups and authority must share one controlled perimeter.
Move from a keyword to an inspectable engineering decision
Bound
Define network, organizational and legal boundaries for the realization.
Inventory
Identify models, APIs, telemetry, registries, caches and bootstrap dependencies.
Isolate
Replace or encapsulate dependencies that violate the selected perimeter.
Challenge
Run no-network, service-loss and complete-restoration tests.
A claim is only as strong as its binding
The gate must evaluate the exact output produced by the exact realization. Integrity, scientific validity and authority are related, but never interchangeable.
Questions that change the design
What does private ai without cloud apis: what must be isolated require?
Removing the generation API is insufficient if embeddings, observability or updates still leave the declared perimeter. Privacy and sovereignty are properties of the executed system: network flows, data, telemetry, dependencies, updates, backups and authority must share one controlled perimeter.
Which evidence should be retained?
At minimum: exact realization identity, inputs and outputs, protocol version, measurements, failure cases and the decision rule. An outbound-flow inventory and negative network test should support every no-cloud-API claim.
Does AMF prescribe a particular model?
No. AMF treats the model, runtime, data, tools and deterministic components as candidate materials. Their retention depends on measured contribution and declared constraints.