AI Knowledge Search for a Public Sector Institution

The Challenge

At the institution, the information that operations, support and knowledge management teams relied on was spread across past records, documents and other scattered sources. When a question came up about a particular topic, record or process, reaching the right information quickly was not straightforward, and an answer could depend on searching several sources separately.

Just as important was what the sources did not contain. Gaps in records, or fields that still needed to be completed, were easy to overlook, which affected record quality and made follow-up harder to manage. The institution needed a way to reach the right information quickly, make better use of what its records already held and notice where something was missing.

The Solution

With cognipeer Studio, our team configured an AI search assistant covering the institution’s past records, documents and other information sources. Operations, support and knowledge management teams, as well as expert users, could ask a question about a specific topic, record or process and receive an answer drawn from the relevant sources. The aim was to make better use of institutional memory, with the teams themselves deciding how to act on what they found.

The assistant scanned those sources and summarised its findings for the user, so they did not need to piece the answer together from separate places. Where information was missing or a field still needed to be completed, it made that visible alongside the summary, so the responsible team could see what needed attention and decide how to follow it up.

Implementation

1. Brought the sources together. We connected the past records, documents and other information sources that teams needed to search, so that a single question could draw on the relevant ones.
2. Configured question-based search. Users could ask about a topic, record or process, and the assistant scanned the relevant sources for an answer.
3. Added summarised findings. Results were presented as a concise summary of what the sources contained on the subject.
4. Surfaced missing information. The assistant highlighted gaps and fields that needed to be completed, leaving follow-up and completion with the relevant teams.

Business Impact

Teams gained a more direct route to the institution’s knowledge, together with a clearer view of where that knowledge was incomplete. Information already held in past records was easier to put to use, and missing details became visible alongside the answer, so teams could act on them through their own follow-up processes.

• Less time spent finding information across past records, documents and other sources.
• More effective use of institutional memory held in earlier records and documents.
• Improved record quality as gaps became easier to see.
• Better-organised follow-up on incomplete records and open points.

Technology Behind the Solution