Home / Notes / Sovereign AI as an engineering property
Notes
What sovereign AI means in practice for our work.
For our work, sovereignty is not a badge. It is the ability to keep data, models, evidence and operating choices under the control of the people responsible for them.
22 September 2026 · Curator Research
01
Alignment, not affiliation
The IndiaAI Mission publicly names indigenous models, India-specific data, application development, startup support and Safe & Trusted AI among its priorities. Curator Research is an independent company working on related engineering goals: Indian context, trustworthy AI, local control and useful applications.
02
What we mean by sovereignty
We build sovereignty through practical controls: knowing where data goes, choosing replaceable model paths, operating locally or air-gapped where the workload and evidence support it, preserving a reviewable record, and remaining economically operable after a grant or promotional credit ends.
03
Where our work fits
Our initial knowledge work is centred on Indian regulatory material. Patra's document core is built around verifiable no-egress execution; the private knowledge and retrieval path is being developed with local operation as a design target. We select open-weight and self-hosted model paths where they are rights-cleared and fit the task, within a mixed technology supply chain.
04
Language access
Patra's approved multilingual architecture is designed for English plus all 22 scheduled Indian languages, with review tiers before a locale is allowed to ship. This website is currently available in English; each additional language would be generated before publication, committed and reviewed as a fixed artefact, with no visitor-runtime translation API. Sarvam Translate is one prospective drafting tool because its current documentation covers all 22 scheduled languages.
05
Current scope
Our present engineering uses a mixed technology supply chain. Support for all 22 scheduled Indian languages is an architecture and design target, with product availability dependent on implementation and locale review. Our public-interest focus is building professional AI systems in India with local control, language access, evidence, safety and affordability as explicit engineering requirements.
Disagree with us? Tell us why.
We publish the measurement, its date and its scope so you can check our reasoning. If you think we got something wrong, write to us.