Glossary
What is Sovereign AI?
Verified 21 September 2026
The term usually refers to compute and models: national data centres and foundation models trained or hosted at home. Less often discussed is the information layer beneath them. An agent built on a sovereign model still needs current, provenance-preserving evidence from the web, and if it obtains that through a search provider abroad, every query — the operator's intent, verbatim — leaves the jurisdiction with it.
A complete sovereign stack therefore has three layers under the customer's control: the compute, the model, and the retrieval and evidence layer that grounds it. The third is the one most often missing.
unlob supplies the third layer as the sovereign edition. It does not supply a model, and it does not decide what is true; it records what was reported, by whom, and what corroborates it, and leaves the judgement to the customer's model.
How unlob handles this: Sovereign edition
Common questions
What is Sovereign AI?
Sovereign AI is the capability of a state or organisation to run AI systems on compute, models and data it controls within its own jurisdiction, without depending on foreign providers.
How does Sovereign AI work in practice?
A complete sovereign stack therefore has three layers under the customer's control: the compute, the model, and the retrieval and evidence layer that grounds it. The third is the one most often missing.
See it working
The free tier is 10,000 credits a month with no card — enough to test any of this against your own queries.