Before someone decides, someone has to search
A case spans messages, contracts and earlier work. Handling often begins with a colleague gathering all of that information.
We developed internal AI research for staff in the FRED environment. It reads the permitted sources and prepares an overview with evidence and unanswered questions. It does not send a reply to customers during that research.
Identify the right case first
Research needs an unambiguous relationship. If a match is uncertain or a source is unavailable, a partial record must not become a supposedly complete case history.
Illustrative example: a colleague wants to understand why a case remains open. AI assembles available information and distinguishes supported observations from questions a person must resolve. This example is fictional.

A useful report also shows what is unknown
| Report component | Purpose |
|---|---|
| Summary | Understand the history |
| Supported observations | Review relevant contradictions |
| Open questions | Avoid treating uncertainty as fact |
| Source gaps | Show limits of the retrieval |
AI can consult further available details in response to a follow-up. The report does not replace a human decision or a missing source.

Research for people, developed with AI
We used AI to build information gathering, limited access and the reporting workflow. This supports internal case handling. It differs from our AI-coordinated field assignments: here the system gathers existing information instead of sending people into the field.
Sources.
- SolvraONE: own development and technical source reviewOwn source · As of



