˅
˅
˅
˅
The GP receptionist has traditionally been the first point of contact between a patient and their GP. This first impression starts to shape the experience of the healthcare journey, before the person even gets to see their Doctor.
Contact with a GP surgery has long been a point of frustration for patients. Receptionists at GP Practices often manage phone lines, walk in enquiries and administrative work at the same time. During busy times it is not unusual for people contacting their GP surgery to experience the engaged tone, being placed on hold or being asked a series of questions whilst receptionists seek to direct patients queries in the right way.
An NHS South East study suggests that 2 in 3 GP practices in Kent and Medway are utilising some form of AI or automated applications that support patient contact.
As more GP practices turn to AI receptionists to address some of the challenges of administrative workload and answer multiple calls at the same time, we wanted to explore if people’s attitudes towards, and experiences of, real and virtual receptionists differed.
NHS guidelines around use of AI tend to focus on implementation of software, compliance with standards and mitigation of risks for GP practices. There is little focus on how AI can impact patient experience.
NHSE guidance to GP Practices, states that practices should ensure that there are alternatives to AI and automated ways for patients to contact the surgery. This might be hugely relevant to vulnerable people, people with disabilities, language barriers or low digital skills and confidence. There is a body of evidence that highlights the risk digital barriers increasing health inequalities.
In June 2026, we spoke to over 700 residents of Kent and Medway. The most frequent frustration was of not being able to get a GP appointment when people felt they needed it. This includes people talking about ‘you have to ring up at 8am to try and get one’. This quote from a person in Medway highlights that the historical frustrations are still being experienced:
Sometimes it can be quite difficult to access GP services as you’re left on hold for a long time, and even sometimes can’t get through to be able to book an appointment. There’s also a big disparity between GP services – for example, some offer online appointment booking and some don’t. If this could be standardised and all GP’s have the online booking system, it could potentially free up the phone lines for the more urgent requests.
It appears that providing an effective receptionist function at GP practices remains a challenge. The patient experience is mixed; some people finding automation enables quicker access to appointments, without the frustration of the engaged tone or being placed on hold.
But the historical themes of negative experience about human receptionists are now being seen in feedback about AI reception tools.
When English is not your first language, when you are not well it is very difficult to explain your condition as something gets lost in translation. Receptionist are triaging you before you get to see a doctor, they should be better trained.
My health conditions can be described as an unsteady stack of plates; each condition (plate) is unrelated to the next, but they are part of that structure and each needs to be considered. Taking one plate out causes the whole stack (me) to collapse. All of this is complex enough that people struggle to understand; a bot has no way of knowing how to use the information I provide it and ends up shutting down.
There are also some newly emerging frustration, in that the automated process is not always ‘turned on’ and that people feel there is no alternative option during those periods. These negative experiences relate to both interactions with AI tools and humans.
I heard about someone who had not sought medical advice recently because they couldn't get an appointment at GP. They phoned reception and were directed to use anima but couldn't access it. They were told that reception was open at key times and tried to call back without success on several occasions. When they did get through to receptionist, they were told to get a neighbour to complete anima for them, but they didn't want to tell the neighbour about their health concerns. So they didn't get advice and ended up in hospital as the situation worsened.
A kind of automated system would be excellent considering I'm profoundly deaf and can't call a reception or front desk for assistance. Yet the bots I've used have left me tired, frustrated, and unsupported.
In October 65% of our Voices Connect panel reported that they had not yet used an AI receptionist, but of those that had, 26% felt that the AI receptionist did not understand the caller's requests.
The learning from this appears that it’s not that people don’t like AI receptionists per se, but rather that they don’t like bad reception experiences, regardless of whether the receptionist is a human or an AI application.
This focus for introduction of AI receptions should be around building people’s trust that the AI receptionist is able to understand their requests. Public trust is essential for the well-functioning of healthcare systems and recent studies have determined that a lack of trust is a structural determinant of health inequity.
We are seeing that there is huge interest in development of AI from both the ICB's perspective as AI starts to demonstrate that it is releasing time that translates to financial savings within Primary Care, but also from the patient's perspectives as to how AI can potentially give a better patient experience that the current human systems.
I support technology, including AI, when it improves communication, accessibility, choice and independence. But technology will only work well if accessibility and lived experience are built in from the start. Health and social care need to work together with disabled people as equal partners, remove barriers and create services around people's real lives. That is how technology, including AI, can help people live gloriously ordinary lives rather than creating another barrier.
