Government

Government teams sit on large volumes of documents, reports, and operational data. This template shows an assistant that turns that into answers and dashboards: staff and reviewers ask questions in natural language and get grounded, cited responses.

How it works

1. Channels – government staff and policy reviewers work through a web client.
2. Gateway & access – a single entry point handles access and authentication.
3. Document ingestion – uploaded documents are stored and embedded so they become searchable.
4. AI agent (+ skills) – reasons about a request and calls its skills: the model for language, retrieval for grounding in real documents, and a dashboard skill for visual answers.
5. NAISH Hub – serves the models on GPU and pulls foundation models from a model hub. Sovereign AI core.
6. Data & dependencies – a vector store grounds retrieval, an analytics store backs dashboards, and identity / SSO gates access.

Where the Hub adds value

• Document intelligence – the Hub-served model reads, compares, and summarises policy and reports at scale.
• Grounding – retrieval keeps answers tied to the government's own documents, with citations, not model guesswork.
• Sovereignty – sensitive government data and models stay on national infrastructure; see Governance & Risk.

Reusable building blocks