Switching AI Scribes in 2026: How to Change Vendor Without Losing Your Notes

If your AI scribe has become unreliable, support has gone quiet, or the pricing no longer fits your team, here is the reassuring answer up front: switching an A...

July 31, 2026

MobileApp

If your AI scribe has become unreliable, support has gone quiet, or the pricing no longer fits your team, here is the reassuring answer up front: switching an AI scribe is far less painful than switching a practice management system. Your patient records live in your PMS, not in the scribe. What you actually migrate is a handful of templates, some workflow habits, and any transcripts or notes still sitting in the old tool. Done carefully, a switch takes about a week of parallel running rather than a data-migration project, and it usually costs nothing beyond one month of overlapping subscriptions. This guide covers what to check before you cancel, how to evaluate the replacement, and the step-by-step playbook that keeps every note safe across the changeover.

One declared interest before we start: Motics makes an AI scribe, so we benefit when clinics switch to us. We have tried to write this guide so it is useful even if you end up switching to someone else, and it includes an honest section on when switching is the wrong move entirely.

The pre-switch checklist

Before you cancel anything, work through this table. It is the part of the process people skip, and it is where the only genuine risks live.

Check before you cancel

Why it matters

What to do

Can you export your templates, or at least copy their text?

Templates carry months of accumulated tuning: your phrasing, structure and specialty quirks. They are the real asset you are migrating.

Export if the vendor supports it; otherwise paste each template's instructions into a document, including any per-section notes.

What happens to retained transcripts and notes after cancellation?

Retention and deletion terms vary widely between vendors. Some delete data promptly after account closure; others retain it for a stated period.

Read the vendor's data-retention terms, and export or download anything you may need before the account closes.

Are any notes still only in the scribe, not in the PMS?

The clinical record belongs in the clinical system. A note that exists only in a scribe you are about to cancel is a governance problem waiting to happen.

Sweep the last few weeks of sessions and push or copy every finalised note into the PMS before you cancel.

When does your billing period end?

The period end sets your natural overlap window: keep both tools live across one billing boundary rather than cold-cutting.

Diary the renewal date and plan the parallel run so the old subscription lapses only after the new tool is proven.

Are individual staff on personal subscriptions?

Clinicians who signed up personally create stray costs and a data-governance mess: patient-related data processed on accounts the clinic does not control.

List every personal account, migrate those users onto the clinic's new plan, and confirm each personal subscription is cancelled and its data handled per the retention terms.

If every row of that table comes back clean, the actual switch is straightforward. The rest of this article explains each step in detail.

Why do clinics switch AI scribes?

The AI scribe market matured quickly, and 2026 is the year the first big wave of switchers arrived: clinics that adopted a scribe in 2023 or 2024 and are now re-evaluating with real usage behind them. The reasons they give follow a small number of patterns. We describe them here as category-wide failure modes, because they are: any vendor, including us, can suffer every one of them.

  • Reliability regressions after product updates. A scribe that worked well for a year starts producing worse notes after a model or feature update. The clinic did not change anything; the output did. Without a change log or status page, it is hard even to confirm that an update happened.

  • Support that cannot help during your working hours. This is the single most common complaint we hear from switchers. Several popular scribes are built by teams headquartered outside the UK, and a support desk that answers overnight, or only by email, is little use when notes are failing during a Tuesday morning clinic.

  • Integrations that break silently. A PMS update, or a change on the vendor's side, stops notes pushing through without any alert. The failure is discovered days later as a backlog of unpushed notes.

  • Per-seat pricing that punishes real clinic staffing. Most clinics are a mix of full-time clinicians, part-timers and associates who do a handful of sessions a week. Per-seat pricing charges the same for each of them, so the clinic ends up paying full price for seats that are used two mornings a week.

  • Governance drift from personal accounts. Individual clinicians trial a scribe on a personal subscription, it sticks, and a year later the clinic discovers patient-related data flowing through accounts it never procured, never risk-assessed and cannot audit.

To be fair to the market: Heidi is the best-known AI scribe in this space and a capable product with a large user base, and Tandem, Lyrebird, Suki, Freed, Abridge and Motics all compete seriously. None of the patterns above is an accusation against any named vendor; they are the recurring shapes of dissatisfaction across the category. We compete with Heidi in more than one category and have published a direct comparison of our phone agent and Heidi Comms if you want to see how we handle a head-to-head fairly.

Do you lose your notes when you switch AI scribes?

No, provided your workflow was sound to begin with. This is the fear that keeps clinics on vendors they have outgrown, and it rests on a misunderstanding of where the clinical record lives.

An AI scribe is a drafting layer. It listens to the consultation, produces a note, and that note is then pushed or copied into your practice management system, which is the system of record. Your retention obligations under the Records Management Code of Practice for Health and Social Care attach to the clinical record in your clinical system, and switching scribe vendors does not touch it. In this respect a scribe switch is closer to changing your email client than changing your filing cabinet.

There are two genuine risk cases, and both are avoidable:

  1. Notes stranded in the scribe. If clinicians have been finalising notes inside the scribe without pushing them to the PMS, those notes exist only in a system you are about to cancel. The pre-switch sweep in the checklist above exists precisely for this. Do it before anything else.

  2. Retained transcripts you still want. Vendors differ on what they keep and for how long after cancellation. Some retain transcripts for a period; others delete promptly. Neither policy is wrong, but you need to know which applies to you, in writing, before the account closes. As an illustration of how much design philosophy varies: Motics deletes consultation audio within 48 hours as standard, and locks the verbatim transcript as a medico-legal record while keeping the generated note fully editable. Other vendors make different, equally defensible choices. The point is to check yours.

Under UK GDPR you also have the right to obtain a copy of personal data a processor holds, and a reputable vendor will handle an export request without drama. If a vendor makes leaving difficult, that tells you something useful, if a little late.

Before you switch, try this

Sometimes the fix is configuration rather than a new vendor, and it is worth an honest hour of diagnosis before you start a migration. Three checks resolve a surprising share of "our scribe got worse" complaints:

  • Re-read your template instructions. Templates work like a recipe: the instructions tell the AI what to include, exclude and how to format. Vague instructions produce inconsistent notes, and a template that worked on an older model version may need its instructions tightened for a newer one. Rewrite the worst-performing section's instructions to be concrete and specific, and re-test before blaming the engine.

  • Check the audio input. Accuracy complaints are very often audio problems in disguise: the device fell back to a built-in laptop mic, the clinician works across the room at a treatment couch, or a new room has worse acoustics. We have written a full guide to microphones and audio setup for clinical AI, and a £49 clip-on mic fixes more scribe problems than any subscription change.

  • Check whether you are running beta features. Real-time transcription modes, experimental note styles and other beta toggles are the most common source of instability in otherwise solid products. If a beta mode is on, turn it off, run a week on the stable path, and see whether the problem follows.

If you have done all three, raised the issue with support using one specific, reproducible example, and the problem persists for weeks with no committed fix, then you have a vendor problem rather than a configuration problem, and switching is reasonable.

How to choose the replacement

Switchers consistently tell us the same thing: the criteria they use the second time are the ones they wish they had used the first time. First-time buyers evaluate note quality on a demo. Second-time buyers evaluate the things that made them leave.

Criterion

What to ask the vendor

What good looks like

Support

"What are your support hours and stated response times? Put them in writing."

UK-hours human support with committed response times, not a best-effort email queue in another time zone.

Transparency about change

"Do you publish a status page and a change log?"

Yes to both, so product updates are visible rather than silent, and regressions can be tied to a cause.

Integration depth

"How exactly do notes reach my PMS, and what happens when the push fails?"

A named, supported integration with your system (not a generic promise), visible failure states, and a workable fallback where no open API exists.

Pricing model

"How do you charge for part-time and associate clinicians?"

Credit-based pricing, or a shared pool, so a clinician doing four sessions a week does not cost the same as a full-timer.

Trial realism

"Can I trial with my own templates on my own consultation types?"

A trial that reproduces your actual workload, not a demo on the vendor's happy path.

Exit terms

"What is your money-back window, and what happens to my data on cancellation?"

A stated money-back period (Motics offers 30 days) and retention terms you can read before you sign.

Two of these deserve a little expansion.

Trial realism is the one that most reliably predicts satisfaction. A scribe demo on a scripted consultation tells you almost nothing; every serious product looks good on its own demo. Rebuild your three most-used templates in the trial account and run them against your real consultation types. If the vendor offers a sandbox for testing template changes against past sessions (Motics does, via the template playground), use it: it lets you validate output quality before a single live patient session depends on the new tool.

Pricing model matters more than headline price for most clinics, because clinic staffing is lumpy. A five-clinician clinic where two are part-time associates pays for five seats under per-seat pricing but consumes perhaps three clinicians' worth of usage. Ask for the pricing to be modelled on your actual rota, and remember that the point of the whole exercise is time: an agentic scribe should be giving each clinician back somewhere around two hours a day, and a pricing dispute over one seat is small next to that.

How long does switching take?

A careful switch takes two to three weeks of elapsed time, of which the active work is a few hours. The structure that works, across every migration we have seen, is a short parallel run that only retires the old tool once the new one has carried real clinical load.

Phase

What you do

You move on when

Week 0: sandbox

Rebuild your top three templates in the new tool. Test them against past or role-played sessions in a sandbox, not live patients. Tighten instructions until output matches your standard.

The new tool's notes on test material need only light edits.

Week 1: parallel run

One clinician uses the new scribe on live sessions, starting with follow-ups (lower documentation risk than new-patient assessments). Old subscription stays live as the safety net.

A full week of real notes has pushed cleanly to the PMS with no workflow surprises.

Week 2: team rollout

The rest of the team moves over, with templates already proven. Old subscription still live but idle.

Every clinician has completed real sessions on the new tool.

Period end: cancel

Complete the pre-switch checklist sweep (stranded notes, exports, personal accounts), then cancel the old subscription at the billing boundary.

Nothing depends on the old tool.

Two principles underneath the table. First, keep both subscriptions live across one billing boundary rather than cold-cutting; the overlap month is cheap insurance and removes all time pressure from the migration. Second, start the parallel run on follow-up appointments rather than complex new-patient assessments, so the new tool earns trust on lower-stakes documentation first.

Note what is absent from the table: there is no data-migration project and no downtime. The patient record never moves because it was never in the scribe.

What does switching actually cost?

Usually nothing beyond the overlap month, and sometimes not even that if your old subscription is monthly and the timing lines up. The costs that do exist:

  • The overlap subscription. One month of paying for both tools. For most scribes this is tens of pounds, not hundreds.

  • A few hours of template rebuilding. There is no standard template-exchange format in this market, so expect to paste template text across and adapt it rather than import it. Budget two to three hours for your top templates.

  • A short adoption dip. The first week on any new tool is slower while habits re-form. The parallel-run structure keeps this contained to one clinician at a time.

For context, Motics pricing (all ex-VAT) runs: a Free plan at £0 with 25 credits a month for a single clinician, no card required, which is enough to trial note quality on roughly 25 real consultations; Starter at £19/month for a single clinician with 110 credits; Team, the most popular plan, from £98/month with 704 shared credits at two clinicians, the Phone agent included, and unlimited users drawing on one shared clinic credit pool; and Scale from £290/month with 2,640 shared credits at two clinicians plus the Audit agent and usage analytics. One credit is roughly one Scribe note, credits scale with your count of full-time clinicians, and unused credits roll over up to 10%. Annual billing saves 15%.

For clinics switching because per-seat pricing punished their part-timers, the unlimited-users structure is often the decisive difference: the plan sizes your credits, not your seats, so an associate who does four sessions a week simply draws a little from the shared pool rather than occupying a full-price seat. In daily terms, Team works out around 16 Scribe notes per clinician per day.

When is switching the wrong move?

An honest guide has to include this section, because a meaningful share of would-be switchers are about to spend effort solving the wrong problem.

Switching is probably the wrong move when:

  • The problem is configuration or audio. If you have not done the three checks in "Before you switch, try this" above, do them first. A new vendor with the same vague templates and the same laptop mic will produce the same complaints.

  • You are inside the learning curve. Satisfaction with any scribe dips in weeks two to four as the novelty fades and the edge cases surface, then recovers as templates get tuned. Switching during the dip just restarts the dip elsewhere.

  • The complaint is one bad week, not a pattern. Every product has bad weeks. A pattern is a fault that persists across a month, that support cannot reproduce or will not commit to fixing. Switch on patterns, not incidents.

  • The vendor has acknowledged the issue with a dated fix. A vendor that says "known regression, fix shipping on this date" and then ships it is behaving well. That is the vendor relationship you want; do not discard it over one regression.

  • Your real problem is elsewhere in the clinic. If notes are fine but the front desk is drowning, a different scribe changes nothing. Diagnose before you migrate.

The clean test: write down the specific failure, when it started, and what the vendor said when you reported it. If the answer to the last part is "nothing useful, for weeks", switch. If you have not reported it yet, report it first.

FAQ

Do you lose patient notes when you change AI scribe vendors? No. Patient records live in your practice management system, which is untouched by a scribe switch. The only at-risk items are notes never pushed to the PMS and transcripts retained inside the old scribe, both of which the pre-switch checklist catches.

Can I run two AI scribes at the same time? Yes, and you should, briefly. The recommended migration keeps both subscriptions live across one billing boundary, with each consultation recorded by one tool only: the new scribe carries live sessions while the old one sits idle as a safety net.

Can I import my templates from my old scribe into the new one? Not directly; there is no standard template format across vendors. In practice you copy the template text out, paste it into the new tool and adapt the instructions. Budget two to three hours for your top templates, and test them in a sandbox before live use.

What happens to my transcripts after I cancel? It depends on the vendor's retention and deletion terms, which you should read before cancelling, not after. Export anything you may need first. Policies genuinely differ: some vendors retain data for a stated period, while Motics, for example, deletes consultation audio within 48 hours as standard.

Do I need to redo consent or paperwork when I switch? Recording consent operates per consultation, so day-to-day practice is unchanged. Administratively, a new vendor is a new data processor, so update your privacy notice and records of processing, put a data-processing agreement in place, and revisit your DPIA. It is an afternoon of governance work, not a project.

When in the year is the best time to switch? Any time your billing dates allow an overlap month, ideally in a normal clinical week rather than your busiest period. Avoid switching in the same month as a PMS upgrade or a new-staff intake; one moving part at a time.

References

  • Records Management Code of Practice for Health and Social Care, NHS England.

  • UK GDPR and the Data Protection Act 2018: rights of access and the storage limitation principle. Information Commissioner's Office guidance at ico.org.uk.

  • Medical device registration in the UK: Medicines and Healthcare products Regulatory Agency (MHRA).


Motics is the AI operating system for clinics: the Scribe agent for consultation notes, the Phone agent for calls, the Audit agent for note-quality review and the Chat agent for working with patient data securely, on a shared credit pool with unlimited users. If you are weighing up a switch, the Free plan is a no-card way to test note quality on your own templates, and every paid plan carries a 30-day money-back guarantee. See how it fits your clinic at motics.ai.

Seven days free. No card. No setup fee.

Up and running in under ten minutes on the free trial. Or book a twenty-minute demo first.

Start free trialBook a 20-minute demo

Used in clinics worldwide · ISO 27001 · HIPAA · GDPR