Reception staffing

Adding AI to the front desk: managing the change

Introduce AI at reception with staff on side: involving the team, redefining roles, patient messages, governance checks and measuring the effect.

Ava research teamPublished 6 min read

On this page
  1. Why the people side decides success
  2. What staff and clinicians worry about
  3. Involve the team before you choose
  4. Redefine roles in writing
  5. Governance checks before go-live
  6. Tell patients plainly
  7. Handling resistance
  8. Measure the effect
  9. Example: a cautious start
  10. Frequently asked questions
  11. References

Key takeaways

  • The technology is rarely what makes an AI rollout fail; unclear roles, rushed communication and missing governance are.
  • Clinicians' worries are specific: among GPs not using AI, 89% cite professional liability, 88% lack of regulatory oversight and 82% patient privacy[1].
  • Involve reception staff in choosing which tasks the tool handles, and write down what each person's job looks like afterwards.
  • Check governance before go-live: data protection impact assessment, clinical safety, medical device status[2] and a human escalation route.
  • Measure a baseline first, then track access, staff workload and patient feedback for at least three months.

Why the people side decides success

Practices that introduce an AI tool at reception, whether for answering calls, replying to messages or sorting online requests, usually spend most of their planning time on the product. The rollouts that struggle tend to do so for human reasons: receptionists who fear for their jobs and quietly route patients around the tool, clinicians who were not consulted and refuse to trust its handoffs, patients who were never told what to expect. Managing those reactions is the main work.

Adoption is still early. In a 2025 Nuffield Trust and RCGP survey of 2,108 GPs, 28% said they currently use AI tools in clinical practice[1]. Most practice teams therefore have little first-hand experience of AI, and their views are shaped by headlines. That makes clear, honest communication even more important.

What staff and clinicians worry about

The concerns are concrete. Among GPs who were not using AI, professional liability and medico-legal issues worried 89%, lack of regulatory oversight 88% and patient privacy and data security 82%[1]. Reception staff add their own questions: will this replace me, will I be blamed when it gets something wrong, and will patients be angrier when they finally reach a person?

Concerns among GPs not using AI toolsNuffield Trust and RCGP survey, 2025
Concerns among GPs not using AI toolsBar chart. Professional liability and medico-legal issues: 89%; Lack of regulatory oversight: 88%; Patient privacy and data security: 82%.Professional liability and medico-legal issuesProfessional liability and medico-legal issues: 89%89%Lack of regulatory oversightLack of regulatory oversight: 88%88%Patient privacy and data securityPatient privacy and data security: 82%82%Concerns among GPs not using AI toolsBar chart. Professional liability and medico-legal issues: 89%; Lack of regulatory oversight: 88%; Patient privacy and data security: 82%.Professional liability and medico-legalissuesProfessional liability and medico-legal issues: 89%89%Lack of regulatory oversightLack of regulatory oversight: 88%88%Patient privacy and data securityPatient privacy and data security: 82%82%

Source: Nuffield Trust, 2025[1]

Show the data
ItemValue
Professional liability and medico-legal issues89%
Lack of regulatory oversight88%
Patient privacy and data security82%

The public is cautious too. Healthcare is the area where UK adults most often say AI needs careful management to prevent negative outcomes, at 29%[3], even though 52% expect AI to have a positive impact on healthcare[3]. A practice that explains what the tool does, what it does not do and how to reach a person will meet most of those concerns.

Involve the team before you choose

  1. Form a small working group

    Include two receptionists, a supervisor, a GP, the practice manager and whoever leads on data protection. Meet weekly for a short, focused session.

  2. Map the work first

    List the tasks reception handles, how often and how long each takes. Use a recent workload count if you have one. Mark the tasks that are routine and repetitive, such as opening hours, how to request a fit note or what the online form is for.

  3. Decide the scope together

    Agree which tasks the tool will handle at first, which it will pass to a person and which it will not touch. Start narrow; you can widen the scope later with evidence.

  4. Test it on yourselves

    Let the working group call, message and try to break the tool before any patient uses it. Record every wrong or unclear answer and fix it.

  5. Agree the success measures

    Decide what would make the change worthwhile and what would make you stop, before go-live.

Be specific about scope in every conversation. For example, an AI Employee such as Ava can answer practice-information questions, explain how to get help, signpost urgent help and pass patient matters to reception; any capability a product lists as planned rather than live should stay out of your plan until it is actually available.

Redefine roles in writing

Staff cope with change far better when they can see what their own job will look like afterwards. Write it down for each role, and be honest where hours or duties will change.

Example of how reception tasks might be reassigned when an AI tool is introduced
TaskBeforeAfter
Routine information callsReceptionist answers every callTool answers; receptionist handles anything passed on
Urgent symptomsReceptionist follows red-flag protocolTool signposts urgent help and alerts staff; receptionist follows up
Complex or upset callersWhoever is freeNamed experienced staff, with more time per call
Online request triageReception and clinician listUnchanged at first
Quality checksAd hocSupervisor reviews a sample of tool conversations weekly

If the change will reduce hours, say so early and explain how: natural turnover, fewer agency shifts, redeployment into care navigation or document work. Uncertainty does more damage than bad news delivered honestly.

Governance checks before go-live

  • Data protection impact assessment: health data is special category data, and processing it is prohibited unless a specific condition applies[4]. Record what data the tool processes, where, for how long and on what basis.
  • Medical device status: the MHRA says many software and AI products used in health and social care are regulated as medical devices[2]. Ask the supplier for the product's regulatory status and intended purpose in writing.
  • Clinical safety: complete a clinical risk assessment for the way you will use the tool, led by a clinician, covering what happens if it gives a wrong or incomplete answer.
  • Escalation: a clear route to a person at any point, and an agreed response time for messages passed to staff.
  • Records: how conversations are logged, who can see them and how they link to the patient record where needed.
  • Exit plan: how you would switch the tool off and return to the previous process within a day.

Tell patients plainly

Patients deserve to know when they are talking to an automated system. Update the phone greeting, website and practice leaflet in the same week, using plain language.

  • Phone greeting: 'You are speaking to the practice's automated assistant. It can answer questions about the practice and pass your message to our team. If you need to speak to a person, say reception at any time.'
  • Website: a short section on what the assistant does, what it does not do, how to reach a person and how your information is used.
  • Practice leaflet and waiting area: one line with the same message, for patients who do not use the website.

Handling resistance

Some resistance is healthy. A receptionist who says the tool gives a wrong answer about repeat prescriptions is doing quality control, and should be thanked and listened to. Other resistance is fear, and it shows up as workarounds: telling patients to press a different option, or saying the system does not work before trying it. Treat both with the same approach: bring the concern into the open and deal with the facts.

  • Hold a short weekly slot for the first two months where staff can raise problems with the tool, and show what was fixed.
  • Ask the most sceptical team members to review a sample of conversations; they often become the strongest quality checkers.
  • Keep the promise about roles; if anything changes, tell staff before it happens, not after.
  • Celebrate visible improvements, such as a quieter 8am queue, with the whole team.
  • If the evidence shows the tool is not working for a task, take that task out of scope and say so.

Measure the effect

Collect a baseline for at least two weeks before go-live and repeat the same measures monthly for three months. Share the results with staff and the patient participation group.

  • Calls answered, abandoned and waiting times by hour.
  • Share of contacts handled fully by the tool and share passed to staff.
  • Time for staff to respond to items passed on.
  • Staff overtime, missed breaks and a short monthly staff pulse question on workload.
  • Complaints and compliments mentioning the phone or the assistant.
  • A sample review of conversations for accuracy and tone.

The government's 10 Year Health Plan commits to rolling out digital triage and AI scribes to cut clinical admin[5], so staff will see more tools in coming years. A practice that learns to introduce one well will find the next one easier.

Example: a cautious start

Example (fictional): A practice introduces an automated assistant for practice-information calls only. Two receptionists sit on the working group and test it for a fortnight. Staff are told in writing that no posts will be cut and that freed time will go to care navigation and the online request list. After three months, morning abandoned calls have fallen, receptionists report more time with complex callers, and the weekly sample review has led to a dozen improved answers. The practice then discusses whether to widen the scope.

The practice did not try to automate everything at once. It chose a small scope, involved the people most affected and used evidence to decide the next step.

Frequently asked questions

How do we reassure receptionists that AI will not replace them?

Be honest about what will change. Write down each role's tasks before and after, explain how any freed time will be used, and involve receptionists in choosing and testing the tool.

Do patients have to be told they are speaking to an automated system?

Being open is the right approach. Say so in the phone greeting and on the website, explain what the system can do and give a clear way to reach a person at any point.

What governance does an AI reception tool need?

At minimum, a data protection impact assessment, a clinical risk assessment for how you use it, written confirmation of the product's regulatory status, a human escalation route and an exit plan.

References

  1. Nuffield Trust (2025). How are GPs using AI? Insights from the front line (opens in a new tab) ↑

  2. GOV.UK (Medicines and Healthcare products Regulatory Agency) (2025). Software and artificial intelligence (AI) as a medical device (opens in a new tab) ↑

  3. GOV.UK (Department for Science, Innovation and Technology) (2024). Public attitudes to data and AI: Tracker survey (Wave 4) report (opens in a new tab) ↑

  4. legislation.gov.uk (The National Archives) (2026). UK GDPR Article 9: Processing of special categories of personal data (opens in a new tab) ↑

  5. GOV.UK (Department of Health and Social Care) (2025). Fit for the future: 10 Year Health Plan for England (opens in a new tab) ↑

Book a demo

See Ava handle your calls.

Book a short demo. We’ll walk through your calls and workflow, and agree what Ava should take care of.

Book a demo