How we’re bringing AI into Cura

We’ve chosen three foundational tools as the building blocks of our AI roadmap:

Speech-to-text and structuring are already in development. Summarisation will come later, because we want experience with AI running in Cura before we move on.

Our first AI functionality: the visit plan

The first concrete AI functionality in Cura will be the visit plan. We’ve chosen it deliberately as a defined area where we can start small, learn a lot, and build the foundation for what comes next.

Here’s how it will work in practice:

1. The staff member dictates the visit plan into Cura.

2. AI listens in, cleans the text of noise and filler words, and structures it according to the visit plan’s fields.

3. The staff member reviews the text, makes corrections if anything is missing or needs rephrasing, and approves it.

This is what we call human in the loop. The staff member always has the final say, because a professional is still the one behind the documentation.

We’re launching a pilot of the visit plan together with Aarhus Municipality, our development partner on this functionality, after summer 2026.

We build it together with the municipalities

Working in partnership with our customers is one of the cornerstones at Systematic. Partnership was also the heart of our first experience with AI in a municipal setting: the TALT project (Danish for ‘spoken’).

In TALT, we were the development partner in a two-year innovation collaboration with Copenhagen, Aalborg, and Aarhus municipalities. The goal was to test speech recognition and text classification as support for documentation in elderly care. In the project’s final phase, 11 municipalities helped test early prototypes.

The results pointed in a clear direction:

• 68% of the community nurses who took part would use AI in its current form to support nursing assessments, if the solution were ready for everyday use.*

• 100% would use AI to summarise records before a visit to an unfamiliar resident.*

• Estimates of time saved per conversation land between 5 and 15 minutes. That’s time that can go back into the meeting with the resident.**

The numbers tell us the direction is right. Users want to use AI in elderly care, because it helps them in their work. But one of the most important lessons from TALT isn’t in the numbers. It’s in the approach:

The municipalities know what daily workflows really look like. Staff are the ones who feel whether an AI suggestion hits the mark or misses it. Without that knowledge, we can’t build tools that make a difference.

It also means we work a little differently than we usually do. With standard software, we can say yes or no to whether a requirement is met. With AI, it’s not that simple. Two different answers can both be right, depending on the context. That’s why we need the municipalities’ professional judgement to define what good quality looks like.

The way forward

The first step is the visit plan, and the pilot with Aarhus Municipality will give us the knowledge we need to take the next ones. What works? Where does the quality land? Where else in Cura would AI tools add the most value?

We can only answer those questions together with the municipalities.

Any municipality that could see itself as a pilot partner, large or small, is welcome to contact their service manager at Systematic. We need the breadth, so that the tools we build will work in different everyday realities.

We’ve already had good experiences with TALT, and so have the users. Now we look forward to developing further AI capabilities for Cura in close partnership with our customers.

___

*Fremfærd Sundhed & Ældre, ‘AI i ældreplejen – Vidensprojekt om brugen af kunstig intelligens til styrket dokumentation’, October 2025

**Figures from TALT prototype trials, September 2025. Time estimates from the TALT project’s evaluation forms, ‘Use cases og afprøvning’, Use case 1 (nursing assessment).