Independent pharmacies shorten phone queues by removing the calls that don't need a pharmacist: automatic ready-to-collect texts cut "is it ready?" calls, a phone menu or AI voice assistant answers opening hours, stock and status questions and takes repeat requests, and callbacks replace holding. Start by counting a week of calls by reason.
Most of the saving comes from sending information before patients call, not from answering calls with AI. A patient who gets "your prescription is ready to collect" at 11am has no reason to ring at 2pm. The AI voice part is useful for what is left, but it needs firm limits: any question about a medicine, a symptom, a side effect or whether two things can be taken together goes to a pharmacist, every time.
Count a week of calls before choosing anything
Owners usually know the phones are bad; few know why. A week of tallying tells you which fix will help. Put a sheet by each phone with a column per reason and have whoever answers make a mark, plus a note of anything that took more than five minutes.
Here is an illustrative week at an independent pharmacy averaging about 140 calls a day:
| Call reason | Share of calls | Needs a pharmacist? | Best fix |
|---|---|---|---|
| "Is my prescription ready?" | 30% | No | Ready-to-collect texts |
| Repeat prescription requests | 20% | No, but needs checking | Online requests; voice assistant capture |
| Opening hours, stock, services | 12% | No | Voice assistant or phone menu |
| Delivery questions | 8% | No | Delivery texts; voice assistant |
| Surgeries and other clinicians | 10% | Often | Direct line or priority route |
| Medicine questions and advice | 15% | Yes | Pharmacist, with a callback queue |
| Other | 5% | Varies | Staff |
In this illustration, well over half the calls never needed a pharmacist, and the biggest single block was status calls that a text would have prevented. Your mix will differ, which is exactly why you count first. A pharmacy with a large delivery service will see more delivery calls; one near several surgeries will see more clinician calls.
Five fixes, from cheapest to most involved
1. Tell patients before they ask
Ready-to-collect messages by text are the cheapest fix and often the biggest. Many pharmacy systems can send them when a prescription is marked ready; AI adds little to the sending, but it can write clearer messages and handle replies.
An illustrative message: "Hi, your prescription is ready to collect from [pharmacy]. We're open until 6pm today and 9am-1pm Saturday. Please bring this message or tell us your name and date of birth. Reply STOP to opt out." Notice what it leaves out: the name of the medicine. A text can be read by anyone holding the phone.
Send "we're waiting on stock, expected Thursday" messages too. A patient who knows the delay does not ring to ask about it. If you send these by WhatsApp rather than SMS, check Meta's pricing: from 1 October 2026 WhatsApp Business Platform service replies become chargeable after the first 1,000 a month per business number, and utility templates sent inside the service window are charged too, at rates that depend on the recipient's market. The free WhatsApp Business app works differently from the platform, as WhatsApp Business app versus API explains.
One more WhatsApp point if you plan an automated assistant there: Meta's platform terms bar general-purpose AI assistants but allow bots tied to a specific business task such as orders, bookings or support. A pharmacy bot that checks prescription status and takes repeat requests fits that; one that chats about anything a patient asks does not, and should not in a pharmacy anyway.
2. Move repeat requests off the phone
Online request forms, apps and email take requests without a queue. The AI job here is reading free-text requests and turning them into something staff can match quickly. An illustrative message from a patient: "Can I have my usual blood pressure tablets and the brown inhaler, and the cream from last time please." An AI assistant working inside your system, or on a de-identified copy, can structure it:
Request type: repeat
Items described:
1. "usual blood pressure tablets" - match to repeat list
2. "brown inhaler" - likely a preventer inhaler; match to record
3. "the cream from last time" - check last dispensed topical
Unmatched / ambiguous: item 3 needs checking with the record.
Action: staff to match items and confirm with patient if unclear.The assistant organises; it does not decide. Staff match each item to the patient's record, and anything ambiguous gets a quick call or message back.
3. A phone menu that listens
A traditional menu ("press 1 for...") with seven options makes callers wait through all seven. Menus with speech recognition let callers say why they are calling and route them in one step. Before and after, illustratively:
Before: "Thank you for calling. Press 1 for repeat prescriptions. Press 2 for opening hours. Press 3 for deliveries. Press 4 for the pharmacist. Press 5 for surgeries. Press 6 for anything else. Press 7 to hear these options again."
After: "Thanks for calling [pharmacy]. In a few words, what are you calling about? For example, 'is my prescription ready', 'opening hours', or 'speak to the pharmacist'."
Speech menus are often a feature of your phone provider's system, sometimes at no extra cost, so ask your provider before buying a separate AI product.
4. An AI voice assistant for the routine calls
Pharmacy-specific voice assistants answer calls in natural conversation, handle status questions, opening hours and stock checks, take repeat requests, and transfer anything else. Several vendors sell these, and the ones worth considering say they integrate with pharmacy systems so they can check real prescription status rather than just take messages. Without that link, an assistant can only say "I'll pass that on", which moves the queue rather than shrinking it. For the wider question of what AI can do on a business phone line, see whether AI can answer your business phone.
5. Callbacks instead of holding
For calls that do need a pharmacist, a callback queue keeps the caller's place without keeping them on the line: "All our pharmacists are with patients. Press 1 or say 'call me back', and we'll ring you within about 40 minutes." Many phone systems include this. It turns a queue of people listening to hold music into a list the pharmacist works through between consultations.
Delivery calls: the same trick again
Pharmacies with a delivery service get a steady stream of "when is it coming?" and "has it been sent?" calls, and they respond to the same fix as status calls: tell people first. An illustrative pair of messages, sent from the delivery schedule:
- The day before: "Your delivery from [pharmacy] is booked for tomorrow, between 1pm and 4pm. If no one will be in, reply CHANGE and we'll rearrange."
- On the day: "Your delivery is on its way and should arrive within the hour. The driver will ask for your name before handing it over."
The AI part is handling the replies: "can they leave it with next door?" or "I'm out until 3" go to staff as structured requests rather than a missed call at 12.55. If your delivery app tracks drivers, a voice assistant linked to it can answer "where's my delivery?" without anyone checking a screen. Keep the pharmacy's rules on leaving medicines with neighbours in the assistant's instructions, and have it pass any such request to staff rather than agreeing on the spot.
What an AI voice assistant should and shouldn't say
Write the rules down before any vendor configures anything, and have the responsible pharmacist sign them off. An illustrative set:
OPENING: "You're through to [pharmacy]'s automated assistant. I can
check if a prescription is ready, take a repeat request, or answer
questions about opening times and deliveries. Say 'pharmacist' or
'staff' at any time to speak to a person."
IDENTITY: before discussing any prescription, confirm full name,
date of birth and first line of address. If any doesn't match:
"I'm sorry, I can't confirm details on this call. I'll put you
through to the team."
NEVER:
- name a medicine or condition unless the caller has passed the
identity check and names it first
- answer questions about doses, side effects, interactions,
pregnancy, children's medicines or symptoms
- say a prescription is ready unless the system shows it as ready
today
- discuss controlled drugs beyond "please speak to the pharmacist"
TRANSFER IMMEDIATELY IF the caller mentions: a symptom, feeling
unwell, a reaction, "can I take", "is it safe", an overdose or
double dose, chest pain, breathing problems, or sounds distressed
or confused. If no one can take the call: offer a callback and say
"If this is an emergency, please call emergency services."The detail that matters most is the transfer list. An assistant trying to be helpful will attempt to answer "I've taken two by mistake, is that OK?" if you let it. It must not. The broader case for keeping some pharmacy work away from AI entirely is set out in pharmacy tasks that should never be handed to AI.
A sample call, and what the assistant got wrong
An illustrative call from the first week of a trial:
Assistant: "You're through to the pharmacy's automated assistant..."
Caller: "Hi, is my prescription ready? It's for my mum."
Assistant: "I can check that. Can you give me the patient's full name, date of birth and the first line of their address?"
Caller: [gives details]
Assistant: "Thank you. That prescription is ready to collect. We're open until 6pm today."
The caller arrived at 5pm and one item was missing: the system showed the prescription as ready because most items had been dispensed, with one owing. The assistant had read "ready" from a status that meant "partly ready". The fix was twofold: the vendor changed the status mapping so part-dispensed prescriptions say "most of it is ready, and one item is on order", and the team stopped marking prescriptions ready until every item was in the bag. The error was not the AI's alone; it exposed a status habit that had been confusing staff for years.
The same call also raised a policy question: collecting for someone else. The pharmacy's existing rules on third-party collection applied unchanged; the assistant was told to say "please bring the usual details for collecting on someone's behalf", taken from the pharmacy's own procedure. If your procedures are not written down clearly enough for an AI to follow, writing pharmacy SOPs with AI is a useful first step.
From 140 calls a day to 95: one pharmacy's month
Back to the illustrative pharmacy with about 140 calls a day. Over a month it switched on ready-to-collect and stock-delay texts, moved repeat requests to an online form promoted on every bag, and added a callback option. It trialled an AI voice assistant in the second fortnight.
- Status calls fell from about 42 a day to about 15, mostly from patients who had not opted in to texts.
- Repeat request calls fell from about 28 to about 16 as regular patients moved online.
- Abandoned calls, callers hanging up in the queue, roughly halved once callbacks were offered.
- The voice assistant handled about a third of the remaining routine calls end to end in its trial fortnight and transferred the rest.
Total calls dropped from about 140 to about 95 a day before the voice assistant did anything, which is why texts and online requests come first. The pharmacist's view was that the real gain was fewer interruptions during consultations, not the call count itself.
Getting the team and regulars used to the new flow
Some regular patients will dislike the change, especially those who have rung the same number for twenty years and know the staff by name. The team's response matters more than the technology. An illustrative line for the counter when a patient complains: "I'm sorry it felt impersonal. The assistant is there so we can answer the phone faster when you need one of us. If you'd rather skip it, just say 'pharmacist' as soon as it answers and it puts you straight through."
Three things help in the first month. Tell regulars before the change, with a card in bags and a sign at the counter, so the first time they hear the assistant is not a surprise. Give staff the transfer words and the callback promise so they can explain both. And put one person in charge of reading the weekly transcript sample and passing complaints back to the vendor, so problems get fixed rather than grumbled about.
What it costs, and how to estimate it
The cheap fixes are often already paid for. Ready-to-collect texts may be included in your pharmacy system or charged per message. Speech menus and callback queues are features of many business phone systems. Online request forms may come with your pharmacy system or a website plugin.
AI voice assistants are usually priced per minute, per call or as a monthly plan with a call allowance; pharmacy-specific vendors often quote after a demo. To compare quotes, work out your own figure: routine calls per day that the assistant could handle, times average call length, times opening days. For the illustrative pharmacy, 40 routine calls a day at about 90 seconds each over 26 days is roughly 1,560 minutes a month. Ask every vendor to price that volume, and ask what happens to the price in your busiest month.
If you use WhatsApp for any of this, add Meta's charges. Since 1 August 2026, Meta's own Business Agent, which replies to messages automatically, has been charged at $2 per million tokens, which Meta estimates at roughly 4-5 cents a message. Platform message charges come on top from October, as above.
Telling callers they are speaking to AI
Say it in the first sentence, as the illustrative opening does. For pharmacies with patients in the EU this is a legal duty as well as good manners: since August 2026 the EU AI Act has required AI systems that talk to people to make clear they are AI, and callers everywhere tend to trust an assistant more when it is upfront. Always offer a route to a person, and make the words to reach one obvious. Setting up an AI phone agent for after-hours calls covers the same principles for when the pharmacy is closed.
Confidentiality on the phone
A call about a prescription is a conversation about someone's health, and the law in most places protects that kind of information more strictly than an ordinary customer record. Four points to settle with any vendor, in writing: whether calls are recorded and how long recordings and transcripts are kept; whether call data is used to train the vendor's models; where it is processed and stored; and whether the vendor will sign a data-processing agreement. Inside the call, the identity check before any prescription detail is the single most important rule. And keep medicine names out of texts and voicemails entirely, because you never know who else is listening or reading.
Measuring whether the phones got quieter
- Calls per day by reason. Repeat the tally for a week after each change. Status calls should fall first.
- Abandoned calls and average wait. Most phone systems report both.
- Calls handled end to end by the assistant, and a weekly sample of 20 transcripts read by a pharmacist for anything it should have transferred.
- Callbacks missed or late. A promise broken is worse than no promise.
- Complaints about the phones. Count them before and after.
If the numbers stop improving after the first two fixes, look again at the tally. The remaining calls are often the ones that genuinely need a pharmacist, and the aim then is not fewer calls but fewer interruptions: callbacks at set times, and a direct line for surgeries so their calls stop joining the public queue.
Questions pharmacy owners ask about AI on the phones
Will older patients cope with an AI voice assistant?
Many will, especially for simple questions, but some won't, so never make the assistant a wall. Offer a person at any point by saying so or pressing a key, and route anyone who sounds confused or distressed straight to staff. Listen to a sample of calls from older patients in the first month and adjust the wording, pace and handoff points.
Can the AI take repeat prescription requests over the phone?
It can take the request and pass it to your team, and with a proper link to your pharmacy system it can check which items the patient usually has. Keep a person matching free-text requests to the record, because callers describe medicines vaguely. The assistant should read back what it recorded and never guess an item it could not match.
Does a callback queue really stop people hanging up?
It usually cuts abandoned calls, because callers no longer have to hold. It only works if you actually call back within the time you promise. Set the promise to what your team can meet on a busy Monday, not a quiet Wednesday, and count missed callbacks weekly. A broken callback promise annoys people more than a queue.
Further reads
- How to Set Up an AI Receptionist Without Losing Callers — Step-by-step setup for any AI phone assistant.
- How to Measure an AI Receptionist's Return in the First 90 Days — Measure whether the phone assistant pays for itself.
- AI Chatbot Disclosure: What to Tell Customers at the Start of a Chat — How to tell people they are talking to an AI.
- How Much Does a WhatsApp Chatbot Cost? Fees, Tools and Setup — What WhatsApp messaging and bots cost to run.
- How to Set Up Missed Call Text Back for a Service Business — A cheap safety net for calls nobody could answer.
- Cutting Parcel-Tracking Calls With AI Delivery Updates — How couriers cut status calls with proactive updates.
- Does AI Stock Ordering Pay Off for an Independent Pharmacy? — Fewer owed items means fewer stock-delay calls to answer.
- How to Survey Your Staff Before an AI Rollout (With Questions) — Fifteen ready-to-use questions, an invitation email and anonymity settings for surveying a small team before you choose any AI tools.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Meta's WhatsApp Business Platform pricing notice (October 2026 changes) and Business Agent pricing; EU AI Act Article 50 transparency obligations; pharmacy voice AI vendor product pages describing status checks and repeat-request capture.