AI-based support for routines, RiTA- Right information to users

Important results from the project

The intent with RiTA was to test, build knowledge about and demonstrate the potential of AI in the public sector. We have achieved this! We have developed a chatbot for employees with routines as a knowledge base. Employees´ understanding of AI as well as organizational learning has developed. Based on lessons learned with RiTA, Uppsala municipality´s IT has managed to concretize parts of its own mission within AI. UU collaboration has offered research-based knowledge about organizational prerequisites for generative AI and contributed to the employees´ knowledge development.

Expected long term effects

How we design our documents has an impact on how well RiTA can assimilate the information in them. It is central to train users in prompting. In addition to being able to provide correct answers, RiTA needs to be likeable in order to increase continued use. We made an active choice to be transparent with the work we do in the project, both to share our learnings, but also to be open to new learnings. We see a clear connection between the time spent on testing and how good RiTA can be based on our needs. This means both an opportunity and a challenge as time is in short supply.

Approach and implementation

The project application was written by those who then worked on the project. We had a well-thought-out and suitably challenging project idea. We got the go-ahead from the management to implement the project regardless of whether we were granted Vinnova funds or not, so we could start immediately. The project had a clear structure and common plan from day one. The project had a strong anchoring in the management. The project staff had the right skills and great commitment. Dialogue with specialists in law, information management and information security began early on.