AI-based speech recognition in elderly care gets the green light: Systematic project demonstrates significant potential

Speech recognition and AI can reduce the documentation burden for frontline staff in elderly care. A groundbreaking project under the Danish Agency for Digitalisation's AI sandbox – a programme for the responsible development of AI solutions – has just documented this in its final report. The conclusion is clear: the AI solution offers significant practical benefits, and the technology can be used lawfully. The next step is scaling up.

A public-private collaboration between Systematic and three of Denmark's largest municipalities – Copenhagen, Aalborg and Aarhus – has spent two years testing the use of AI-based speech recognition and automated clinical note-writing in elderly care.

The aim was straightforward: can AI and speech recognition free up time spent on documentation, giving healthcare staff more time with residents, without compromising the quality of record-keeping?

The answer is yes.

The final report from the Danish Data Protection Authority and the Agency for Digitalisation shows that the technology saves time, works well in practice, and can be used lawfully within the existing regulatory framework.

Healthcare professionals see the potential

The project, named TALT (meaning ‘Spoken’ in Danish) has also involved extensive user testing and use cases in which nurses participated in trials of the speech-to-text functionality. The feedback has been overwhelmingly positive:

"It frees up time when you can dictate, for example, the daily care plan instead of writing everything down. It also creates more presence with the resident during the conversation, when you don't have to sit and write while you're talking," explains Birgitte Lysén Jensen, nurse at Plejecenter Toftehøjen in Viby Sjælland, in the report Afprøvning og use cases – fra et sundhedsfagligt perspektiv (Project TALT, February 2026).

Fellow nurse Sheela Korami also highlights the importance of presence in the same report:

"It can help create more face-to-face contact with the resident, when you no longer need to document whilst speaking with them."

Nurse Thomas Rasmussen from the community nursing service in Vanløse can also see the benefits of using AI – not least in connection with nursing assessments. However, he stresses that professional judgement must remain central:

"The system produces a draft of the documentation across the nursing problem areas. That could potentially make it easier for us to document. The system is not a replacement for professional expertise – it is a supplement, and it is important that we approach its documentation suggestions with a critical eye," Thomas Rasmussen emphasises in the same report.

Birgitte Lysén Jensen will always review the documentation suggestions to ensure accuracy. But she has no doubt that a fully developed AI solution would be of great help – and that she would be ready to use it tomorrow.

Ready for national roll-out

The project did not start from a finished solution, but from concrete use cases – including one in which a member of staff dictates their notes into a phone, and the system automatically summarises and structures the data in the resident's care record. That scenario clearly resonated with users: 100% of participating staff said they would use AI-generated care record summaries in their daily work, if the functionality were available.

Systematic provided the solution along with technical expertise and insight into how speech recognition and AI can best be integrated into existing health IT systems. According to the report, the technology is now ready for further development and scaling – not least because documentation requirements in elderly care are growing year on year, without a corresponding increase in staffing.

"If we want to retain staff and create more presence in elderly care, we need to embrace technologies like this. But it must be done thoughtfully, and that requires solutions built for the real world – not for the drawing board," says Brian Fisker.

The success of the TALT project has been so compelling that speech recognition was selected at the end of 2025 as one of three national large-scale AI projects by the Danish Agency for Digitalisation, KL (the Local Government Denmark association) and Danish Regions.

"It is a success for the TALT project that speech recognition has become one of the joint public large-scale projects. We have sought to shape the national agenda in that direction, because we could see it working and saving time. We are now in a very strong position to further develop and deploy a solution at scale," says Rikke Saltoft Andersen, Head of Digitalisation at the City of Copenhagen and project manager for TALT.