An internal AI and data platform for a public-sector organization: a document assistant that only answers from official sources, analytics dashboards, and automated weekly reporting. Client name kept confidential.

This client is a public-sector organization whose name we keep confidential. Their institutional knowledge lived in PDFs, and their reporting was scattered across systems and manual work. We built an internal RAG platform where staff upload PDFs and ask questions in plain language. Every answer is generated strictly from the uploaded documents, and when the documents do not contain the answer, the assistant says so instead of guessing. That design keeps hallucinations out of official work. Around it, we migrated their legacy data into a clean database, built Power BI dashboards and data analysis on top, and set up email automations that run SQL queries on a schedule and deliver weekly reports without anyone lifting a finger.

Staff upload PDFs and ask questions in everyday language, getting answers in seconds instead of digging through files. An admin panel lets the team create assistants per topic, keep them updated, and control exactly who sees what.

Uploaded documents are indexed into a vector database, and every answer is generated only from the passages retrieved from them. If the documents do not contain the answer, the assistant says so rather than making something up, and sources stay visible so staff can verify everything.

We migrated scattered legacy data into a clean database, then built Power BI dashboards and analysis on top. Monthly snapshots, year over year trends, and breakdowns that used to take manual spreadsheet work are now a few clicks.

Scheduled automations run SQL queries and email the results out every week. Each report can be toggled, scheduled, and pointed at the right recipients, so recurring reporting no longer takes up anyone's time.