TL;DR
Fakenham Medical Practice in Norfolk is ending its trial of EMMA, the AI phone receptionist made by QuantumLoopAI, after patients and staff complained about how it worked. The practice started using it in April and will switch in October to a “total triage” model run by clinicians. Its chief executive admitted the tool “didn’t work as intended”.
What went wrong
EMMA takes calls during opening hours, logs why the patient is ringing and hands the details to reception staff. QuantumLoopAI says it serves over two million patients at practices across the country. Fakenham brought it in to cope with phone demand, but complaints about the experience and the service built up.
“We appreciate that this will have been incredibly frustrating for many of our patients,” Andrew Havemann, the practice’s chief executive, told the local paper, the Fakenham and Wells Times.
Under the replacement model, a clinician will review and prioritise every patient request before directing it to the right professional. Havemann said the change would be “more consistent and equitable”.
QuantumLoopAI said it respected the decision and took the feedback seriously. “Introducing AI into a busy surgery is a big change for staff and patients and making it work is our job as much as the practice’s,” a spokesperson said.
A pattern, not a one-off
Fakenham is not the first practice to have trouble with the system. A July blog from Healthwatch Rotherham said patients with regional accents struggled to make EMMA understand them, as did people with speech difficulties. Weeks later, a Doncaster surgery withdrew EMMA after a stroke survivor failed repeatedly to book and moved practice. Healthwatch Rotherham is now sending its evidence on to Healthwatch England, which in turn reports to DHSC.
Looking forward
The market is still growing. On the same day, rival supplier InTouchNow announced £2.3m of seed funding for its own GP voice agent. For practice managers weighing these tools, three EMMA stories in five weeks point to the same questions: how the system handles accents and speech impairments, what happens when a caller gives up, and whether there is an easy route to a person. If Healthwatch England’s report reaches DHSC, national guidance on AI telephony in general practice could follow.