Anonymised under NDA
Executive intelligence briefings and chat for an Abu Dhabi government entity
Government & Public Sector · A government entity in Abu Dhabi
A system that pulls data from multiple source systems into one governed store, generates executive presentations automatically from what arrives, and gives leadership a chat service to ask questions of the same data.
The problem
- Executive reporting was assembled by hand from several separate systems, so every briefing took days and was out of date by the time it was presented.
- Each source had its own definitions and formats, and reconciling them was repeated for every reporting cycle.
- Leadership had no way to ask a follow-up question without going back to an analyst and waiting for the next cycle.
The architecture
- Ingestion pipelines pull from multiple operational and reporting sources on a schedule, normalise them against a shared data model, and record lineage for every figure.
- A generation layer turns the current state of the data into executive presentations: narrative, charts, and tables produced from templates and an LLM, with every number traceable to its source.
- A chat service over the same data, using retrieval with source-grounded citations, so executives and their offices can query the figures behind any slide in plain language.
- Deployed inside the entity's estate to meet UAE data-residency and classification requirements, with role-based access and a full audit log of queries and generated outputs.
ClaudeLlamaPostgreSQLQdrantPythonPowerPoint generation
The outcome
- Executive presentations are generated from live data rather than assembled by hand, and are current on the day they are presented.
- Leadership can interrogate the data directly, with a cited answer, instead of waiting for the next reporting cycle.
Services drawn on
A similar problem on your roadmap?
A Discovery Sprint is the fastest path to a scoped, costed proposal.