During an AI trial at St Wulfstan’s Surgery in Warwickshire, one patient noticed something that had little to do with software: the doctor was looking at them instead of the computer.
“Listening rather than typing was very positive,” the patient said.
Clinicians in the trial spent 16.7% of consultations looking at a computer when using Tandem, compared with 74.3% without it. Patients also reported stronger engagement. For a company built around medical documentation, the result captured the original promise neatly: make the computer quieter so the conversation can become louder.
Tandem Health has now raised $100 million in Series B funding, led by the Scaleup Europe Fund managed by EQT, with Kinnevik, Northzone, Amino Collective and Visionaries participating. The round takes total funding to $160 million. Tandem says its technology is used by more than 10,000 care organisations across 14 European markets.
Founded in 2023 by Lukas Saari, Oliver Åstrand and Oscar Boldt-Christmas, Tandem Health began with a deceptively narrow problem: clinicians were spending too much of the consultation documenting what had happened instead of focusing on the person in front of them.
Doctors had previously dictated not only what happened during a consultation, but what needed to happen next: send the referral, arrange another appointment, write the letter, follow up with the patient. When AI took over the note, that informal handoff did not automatically come with it. Clinicians could still find themselves leaving one system, opening another and rebuilding the context for somebody else.
The note was finished. The work inside it was not.
That discovery pushed Tandem beyond transcription. Tandem Secretary keeps the clinical note, delegated task, responsible person and completion update together. Its broader medical assistant now spans preparation, documentation, coding and clinical decision support, while the company plans to extend further into scheduling, triage, patient communication and task management. It is already integrated with more than 130 medical-record systems.
There is evidence for the documentation layer, though it should not be overstated. A study involving editing metadata from 236,153 generated notes found that sustained users estimated documentation time falling from 6.69 to 4.72 minutes per note, roughly 29%. The study was observational and focused on established adopters, so it does not prove the same benefit for every clinician or better patient outcomes.
That makes Tandem’s next test harder than generating good prose.
A clinic does not run on notes alone. It runs on whether the referral was sent, the appointment was booked, the result was reviewed and the patient was contacted when someone said they would be.
Tandem started by teaching AI to capture what happened in the room.
Its $100 million bet is that the harder opportunity begins when the patient walks out.
