A compounding pharmacy should check six things before buying software: formula and batch records, ingredient and scale verification, labelling, refills and reminders, an audit trail, and support. AI is a minor point: the one vendor I found describing pharmacy AI in detail lists prescription pre-fill, claim help and reporting, all still upcoming.
That order matters because a compounding pharmacy is a manufacturer in miniature. A system that is wonderful at dispensing tablets and clumsy at a master formula will cost you more than any feature gets back. Treat the checklist as the buying decision, and the AI section as a short test to run once the foundation passes.
Group one: formulas and batch records
Everything else rests on this. DataScan's compounding page, for example, lists building, copying and modifying master formulas, sub-formulas, integrated batch and lot generation, linked equipment lists, lot and expiration tracking, master formula compound logs, batch compound logs and patient compound summary sheets. PK Software, which belongs to the compounding supplier PCCA, describes e-prescribing, production and batching, labelling, inventory and billing in one platform, with logs and reporting for audits. Whatever the system, each of these items should be shown to you, not described.
| Check | What good looks like | How to verify in a demo |
|---|---|---|
| Master formula build | Ingredients, quantities, method steps, beyond-use dating rules and equipment in one record | Ask them to build a formula you use weekly, live |
| Scaling | Change the batch size and quantities recalculate, with rounding rules you can read | Scale a 30 g cream to a 1 kg batch and inspect the numbers |
| Sub-formulas | A base used inside several formulas, updated in one place | Change the base and see which formulas it touches |
| Batch and lot | Batch number generated, ingredient lots recorded, expiry rules applied | Make a batch and open the batch record |
| Version control | Old versions of a formula retained and visible | Edit a formula, then view the history |
| Linked documents | Safety data and certificates attached to a compound | Attach a document to a compound record |
Group two: ingredient and scale verification
This is where accuracy is protected. DataScan lists ingredient barcode verification and integration with Ohaus and Sartorius scales. Not every system does either, and some vendors partner with a third-party tool. Ask what happens when a weighing is out of tolerance: does the system stop the batch, warn, or let you carry on and write a note? Ask whether the weights are captured into the batch record automatically or typed.
A realistic test: scan the wrong ingredient bottle on purpose. A decent system refuses it. A weak one accepts the scan and you find out at the label stage, or worse, at review.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowGroup three: order intake, labelling and the doctor's side
Compounding orders often arrive from prescribers with unusual requests, so intake needs to be flexible and tidy.
- Prescriber portal. DataScan describes a doctor portal where prescribers log in, see the patient's history and create and track compound prescriptions. Ask any vendor whether a portal exists and whether orders arrive as structured data or as text you must retype.
- E-prescribing. PK Software's site lists e-prescribing in its platform, alongside auto-population of prescriptions, prescribers and patients to cut manual entry. Check that compound prescriptions, not only standard ones, can arrive electronically, and which fields then fill themselves.
- Numbering. PK Software's site mentions flexible numbering and a clear separation of prescription types (controlled, office use and over-the-counter) for record-keeping. A mixed pharmacy needs that separation.
- Labels. Custom stock labels with barcodes, patient labels with beyond-use dates, warnings and storage notes. Print a real label and check every field against your local labelling rules.
- Payment before work. PioneerRx's Click to Pay, announced in November 2023, texts the patient a secure link to pay before an expensive compound is made, using rules the pharmacy sets for when requests go out. If you have had unclaimed expensive orders, ask any vendor for the equivalent.
Group four: refills, reminders and communication
Compounds often have long refill cycles and patients who forget. Check refill reminders by text or voice, how a prescriber's change is recorded, and whether a patient can request a refill without phoning. Ask where those messages come from: the pharmacy system itself or a connected service, since that changes who holds the patient's contact data. A pharmacy with many repeat patients will notice reminder quality first.
Group five: the audit trail and access control
An inspector asks who did what and when, so make the system prove it. Check these.
- Every edit to a formula, batch or prescription is logged with user, time and old and new values.
- Roles separate who may create formulas, who may compound, and who may verify.
- Deleted records can be recovered, or deletion is blocked.
- A report shows all activity for one batch or one patient on demand.
- Exports of records are available in a readable form for inspection or if you leave the vendor.
Group six: support, switching and the contract
DataScan's pages emphasise support response times and training time, and PK Software describes onboarding, live support, training resources and ongoing education led by its customer success team. Treat those as claims to test: ask for two references from pharmacies of your size and call them. PK Software also says it is moving to the cloud through an early adopter programme; when a vendor is mid-move like that, ask which version you would be buying, how long the current version stays supported, and whether the move itself is included in the price. Then check the contract for the term, price increase rules, who owns the data, what a data export costs, and what happens at the end. Also ask for a written migration plan covering master formulas, patient profiles and open refills, with a parallel run before you stop the old system.
Where AI genuinely fits, and where it is marketing
Having passed the six groups, you can look at AI with a calm head. Here is what I could verify from vendor pages.
DataScan has a page for DataScan AI. It describes three features as upcoming: AI-powered prescription filling that reads an incoming e-prescription, matches the patient, prescriber and drug, and pre-populates the fill screen for a technician to review before pharmacist verification; claims rejection assistance that explains a rejection in plain language and suggests next steps while the team stays in control of resubmission; and an AI-assisted report builder that turns a plain-language request into a report. Further items, such as adherence outreach and help for using the software, are labelled in development. The page tells readers to contact the company for current availability and release dates, and its FAQ gives a target of release before the end of 2026 for automatic filling and the report builder, within six months for the other features, and names AI-assisted inventory forecasting as planned. In other words, as of November 2026 the most detailed pharmacy-AI description I found is largely a roadmap.
Listing sites tag some pharmacy systems, PioneerRx among them, with AI features, but those tags do not say what the AI does. PioneerRx's own software page, when I read it, listed compounding tools and integration with compound-specific scales but named no AI feature. That is a reason to ask, not a reason to dismiss.
Useful automation is not always labelled AI, either. PK Software describes auto-population of prescriptions, prescribers and patients as a way to cut manual entry, with no AI claim attached. For a buyer the label matters less than the job: ask what fills itself, from which source, and what the technician still checks, whatever the brochure calls it.
| AI claim type | Plausible value for a compounding pharmacy | Test it with |
|---|---|---|
| Reads e-prescriptions and pre-fills the screen | Saves keystrokes on standard prescriptions; less help with unusual compounds | Feed it three compound prescriptions with free-text instructions |
| Explains rejected claims | Helpful for the billing person on routine rejections | Give it five past rejections and compare with what you know |
| Builds reports from plain language | Useful if your data is clean; saves the afternoon spent building a query | Ask for a report you already run and compare the numbers |
| Chat or voice assistant for refills | Reduces phone queues; needs safe handover | Role-play a refill request with an unusual detail |
| "AI-powered" compounding or formulation | Treat with suspicion unless a pharmacist can see the source and own the decision | Ask what it generates, from what, and who signs it off |
The last row deserves a firm line. A tool that suggests a formula, a dose or a beyond-use date is making a clinical decision. Ask for the evidence source, let a pharmacist check every output, and never let it be the only check. The related list of jobs that should stay with a person is in which pharmacy tasks should never be handed to AI.
Testing an AI pre-fill with prescriptions that are not tidy
A pre-fill feature will look excellent on a clean prescription for a standard drug. Compounding prescriptions are rarely clean. Before you trust any tool that reads incoming prescriptions, run it on the awkward ones you actually receive. Here is an illustrative test set and what a careful reviewer should look for.
| Test prescription (invented) | What a good tool does | What to mark as a fail |
|---|---|---|
| Free text: "Hydrocortisone 2.5% in a cream base, 30 g, apply twice daily, no preservative" | Pre-fills ingredient, strength, quantity and directions; flags "no preservative" for the pharmacist | Drops the no-preservative instruction or picks a standard product |
| Abbreviations: "Cap, 10 mg x 60, 1 qhs, ref x2" | Reads the abbreviations correctly, or flags them as unclear | Guesses silently |
| Two patients with similar names from one prescriber | Asks the technician to confirm the match | Picks the wrong patient |
| Handwritten fax image | Says it cannot read it with confidence | Produces a clean-looking fill from a blurry page |
Count the outcomes over twenty prescriptions: right and pre-filled, right with a flag, wrong and unflagged. The third column is the number that matters. A tool that is wrong but unflagged on even one in fifty needs a firm review rule: the technician checks every pre-filled field against the original, not just the ones that look odd. If your pharmacists would then spend as long as keying it in, the feature does not save time for compounds, though it may for standard prescriptions.
Checking the "explains rejected claims" promise
This feature is easy to evaluate because you already know the answers. Collect five rejections from the last month with their causes: an expired prior authorisation, a quantity limit, a missing diagnosis code, a refill-too-soon, a non-covered compound ingredient. Give each to the tool. Mark whether it names the real cause, offers a sensible next step, and avoids suggesting anything that would be improper, such as altering a field to force acceptance. The vendor's own page says the team stays in control of every resubmission, which is the right design. Your check is whether the explanation is accurate enough that a junior colleague would follow it. If two of five are wrong, the feature can still help a senior biller who knows when to ignore it, but not a new starter.
The first thirty days after you choose
Buying is half the work. Here is a calendar that keeps a switch under control.
- Week one: build. Enter your twenty most-used master formulas and have a second person verify each against the source. Set up users, roles and label templates. No live patients yet.
- Week two: shadow. Run the new system alongside the old for a few days. Compare labels, batch numbers and quantities line by line. Log every difference.
- Week three: go live for standard orders. Keep complex compounds in the old system until the formulas have been proven. Write the AI features policy: which are on, for whom, and the review rule.
- Week four: review. Count errors caught, minutes saved and complaints. Decide whether to move the complex compounds, and set the next review date.
Keep the old system readable for as long as your record-keeping rules require. Ask the old vendor now what an export costs and whether it is possible after the contract ends, because that is easier to negotiate before you tell them you are leaving.
Cost questions that belong on the same page as the checklist
The prices I found for compounding pharmacy software were mostly quote-based, so the sum is yours to build. Ask each vendor to price these lines separately so you can compare:
- Licence per workstation or per pharmacy, and what counts as one.
- Compounding module and any prescriber portal, if separate.
- Scale and barcode hardware, and whether you must buy specific models.
- Support and training, including what is free for the first year.
- Data migration, and whether paper records are scanned or retyped by you.
- Any AI features, whether included, an add-on or "coming soon" with a price to be set.
- Price changes: how much the vendor can raise the fee at renewal.
Putting a vendor's quote into these seven lines often shows that two systems with similar headline prices differ by thousands over three years because of hardware, migration and support. The cheapest line item is rarely the cheapest system.
A scored checklist you can print
Here is a filled-in example for a pharmacy comparing two systems. Score each line 0 (fails), 1 (partial), 2 (shown live and works). The numbers are invented to show the method.
Pharmacy: [name] Date: [date] System A System B
1. Master formula built live from my own recipe 2 1
2. Batch resized and quantities recalculated 2 2
3. Wrong ingredient scan refused 1 2
4. Scale integrated, weights captured automatically 2 0
5. Label printed with all required fields 2 2
6. Prescriber portal or e-prescribing for compounds 1 2
7. Refill reminders by text or voice 2 1
8. Audit trail shows edits with old/new values 2 1
9. Roles separate create / compound / verify 1 1
10. Data export and exit terms in writing 1 2
11. Reference call with a pharmacy of my size 2 1
12. AI claims shown live, or clearly labelled roadmap 1 0
Total (max 24) 19 15Do not just add up the numbers. A system that scores a 0 on a safety line, such as the audit trail or the wrong-ingredient refusal, should be out regardless of the total. Use the total to break ties among systems that pass the safety lines.
A worked scenario: a small pharmacy replaces paper logs and labels
Illustration. A pharmacy with four staff makes about 40 compounds a day, mostly creams and capsules, and keeps batch logs partly on paper. A tech spends roughly 90 minutes a day on labels and logs, and the pharmacist spends 30 minutes on end-of-day checks of the paper trail.
The owner shortlists two systems and runs the demo script from the checklist above. System A shows everything live. System B shows slides and says the scale integration is "available". The owner asks System B for a reference from a pharmacy of the same size and gets one after a week. The reference confirms the integration works but says setup took three weeks, not the two the vendor quoted.
The owner estimates the saving from System A at 60 minutes a day of tech time, since batch logs write themselves and scales feed weights in. Over 22 working days that is 22 hours a month. The choice is not purely financial, because the biggest gain is fewer transcription errors and a trail that stands up to inspection. But the numbers help to frame the conversation with the supplier. The owner then asks for a written price, migration plan and a trial month before signing.
What goes wrong in a switch, and how to see it coming
- Formula drift. Master formulas are retyped from a paper folder and an ingredient quantity differs by a decimal place. Fix: a second person checks every migrated formula against the source, and the first batch of each is logged as verified.
- Open refills lost. Patients with active refill series go missing in the move. Fix: list them before migration and check each one after.
- Labels that pass the system but fail inspection. A field is missing, such as a storage note. Fix: check the printed label against your regulator's rules, not the vendor's template.
- AI switched on by default. A pre-fill feature turns on for all users. Fix: ask in writing how to switch it off per user and per feature.
Questions to send a vendor before the demo
Subject: Compounding software: demo preparation
Hello [first name],
Before our demo, please send:
1. A list of the compounding features included in the price we discussed, and any that cost extra.
2. Whether scale integration, ingredient barcode verification and a prescriber portal are included, and which scales are supported.
3. A description of any AI features, with a statement for each of: generally available, beta, or in development.
4. Where our data is stored, how long you keep it after we leave, and the export format.
5. The migration plan for master formulas, patient profiles and open refills, and the typical timeline.
6. Two reference pharmacies of similar size we can call.
We will bring three of our own formulas to the demo and would like to build them live.
Kind regards,
[your name]Hold a vendor to the answers. If the demo differs from the email, ask which is correct and put the answer in writing. For a general question set that applies to any pharmacy supplier, see the buyer's checklist for pharmacy automation companies, and for where AI is starting to take phone and refill work, phone, chat or refill requests first sets out the order I would use.
Local rules, and who must sign off
Compounding is regulated, and what must appear on a record, label or beyond-use date differs by place and by the type of preparation. No software checklist can tell you what your own rules require. Take your regulator's current requirements, your professional standards and your own standard operating procedures, and use them as the test for every label and record the system produces. If you need to rewrite procedures for a new system, writing pharmacy SOPs with AI shows a workflow, but a pharmacist must approve every procedure. Software is only ever evidence that you followed your procedures. It is not a substitute for them, and no vendor can sign off your compliance for you.
Questions pharmacists ask when buying compounding software
Can a standard pharmacy system handle compounding?
Some can, but compounding needs master formulas, batch and lot records, ingredient verification and beyond-use dating that dispensing systems may only partly cover. Ask for a live demo of a real formula from creation through to the label and the batch record before assuming the module is enough.
Should AI be a reason to choose one system over another?
Rarely, today. The vendors I checked describe AI mainly for reading e-prescriptions, explaining rejected claims and building reports, and one labels those as upcoming. Choose on formulas, records, labelling and support first, and treat AI as a bonus you test.
Who is responsible if an AI-filled prescription is wrong?
The pharmacy. Whatever a tool pre-fills, a pharmacist verifies the prescription and the compound, and the record shows who did. Check that the system logs both the AI step and the human check, and never allow the tool to skip verification.
How long does switching compounding software take?
It depends on the number of master formulas, patient records and open refills to move. Ask the vendor for a written plan listing what migrates, what you retype and who checks it. Budget for a parallel run of a few days, and keep the old system's records accessible.
Further reads
- How Independent Pharmacies Use AI to Shorten Phone Queues — Shortening phone queues in an independent pharmacy.
- Does AI Stock Ordering Pay Off for an Independent Pharmacy? — Whether AI stock ordering pays for itself.
- Questions to Ask an AI Vendor Before You Sign Anything — A general question list for any AI vendor.
- What to Check in an AI Tool's Privacy Policy and Terms — What to read in an AI tool's privacy terms.
- Pharmacy Staff Training: Induction Packs for Counter Staff With AI — Training staff on a new system with AI-drafted packs.
- How to Evaluate an AI Implementation Proposal or Quote — How to judge a vendor's quote and proposal.
- Best Pharmacy Dispensing Software: Does Any of It Have Real AI? — Six pass-or-fail tests that separate genuine AI from rules-based automation in dispensing systems, with a scored worked example.
- AI Software for a Community Pharmacy: Patient Reminders and Delivery Rounds — Write pick-up reminder templates with AI, send them through the pharmacy system, and plan delivery rounds from stop codes instead of patient names.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: DataScan Pharmacy compounding and DataScan AI pages, PK Software and PCCA product pages, and PioneerRx's software page and Click to Pay announcement (viewed 4 November 2026). Vendor claims are paraphrased; availability dates are the vendors' own.