AI Platform for Doctors: One Suite or Separate Scribe and Phone Tools?

Buy one suite if your practice has roughly five clinicians or fewer, no deep EHR integration to protect, and you want one contract and one BAA. Buy separate scribe and phone tools if your EHR already pairs with a scribe or if either function is too weak in the bundle. Get a BAA from every vendor and run an integration test first.

The question is a buying decision, not a clinical one. It turns on what each option costs to run, how many agreements you must review, what breaks when one vendor fails, and how hard it is to leave. Features come last, because most listed features go unused.

Follow me on Instagram@sagnikteaches

What counts as a suite, and what counts as separate

A suite sells several jobs from one vendor under one login and contract: usually a scribe, a phone receptionist, scheduling, and an inbox for faxes, email and texts. DeepCura describes itself in these terms. Its pricing page bundles an AI scribe, an AI receptionist with a phone number, AI fax, scheduling, code suggestions and EHR integration for $129 a month per provider seat on monthly billing, with seats three to five at $116, or $999 a year per seat on annual billing. Each seat carries 1,000 monthly credits shared across its AI tools. That is a vendor description and price, and worth checking again before you rely on it.

Connect on LinkedInSagnik Bhattacharya

Separate tools are single-purpose products chosen for each job and connected to your EHR or to each other: a scribe such as Freed (Core $79 a month and Premier $104 on its page's default view; Premier shows $119 on the monthly-billing toggle) or Heidi (a free tier, and a Clinician plan at $110 per user a month billed annually), plus a phone assistant from a different vendor. A general AI receptionist such as Smith.ai lists a free plan covering 25 calls a month and then $3 a call, and a Pro plan at $150 for 75 calls, with tiers for 150 and 300 calls (the 300-call tier is about $500). Its AI receptionist pricing page does not mention health data or a BAA, so whether any plan suits a clinical setting is a question to put to it, not an assumption.

Subscribe on YouTube@sagnikteaches

A third group sits between the two: EHR-native tools. Microsoft's Dragon Copilot, for example, is positioned around health systems and EHR partners such as Epic, shows no public price and is bought through a sales contact, so it is enterprise-focused rather than a small-practice purchase. If your EHR vendor offers its own AI scribe or phone feature, count it as a candidate for either side of the comparison.

The five costs that decide it

Read the table by row, not by column. If your practice cannot spare anyone to manage vendors, the agreements and integration rows dominate and the suite wins. If you already have a strong EHR and a clear favourite scribe, the weakest-link and exit rows dominate and separate tools win.

The sums for a four-provider practice

Illustration: four providers, one front-desk team, 500 inbound calls a month, of which AI answers about 300. All prices are list prices taken from vendor pages on 4 November 2026 and need re-checking; the separate stacks use a phone tool priced like Smith.ai's 300-call tier.

Two points. First, on list prices the suite costs $326 a month less than the cheapest separate stack here, and about $480 less on annual billing. That gap is only real if the suite's 1,000 credits per seat cover 300 calls plus every note, so ask the vendor how many credits a call and a note use before you count it. Second, per-call pricing overtakes a tier quickly: on Smith.ai's figures, pay-per-call costs the same as the $150 Pro tier at 75 calls and $825 at 300. If your practice takes 60 calls a month, the separate phone side costs $105, Freed Core plus that comes to $421, and the suite is no longer the cheaper option. Do them with your own call volume before you accept either pricing shape.

Add the staff time

Say managing two vendors takes the office manager two hours a month more than managing one: renewals, logins, a support ticket, a template change. At $35 an hour that is $70 a month, which widens the suite's lead in this illustration to roughly $400 a month. The separate route therefore wins on quality and fit, not price: when the suite's scribe or phone function fails your test, when its credits run short at your volume, or when your EHR already pairs well with a specialist scribe.

A decision rule by practice size and EHR

• Solo or two clinicians, light EHR. Start with one suite, or with a scribe alone, and add a phone tool later. A small practice rarely has anyone to manage several vendors, and at list prices a mixed stack seldom saves money.
• Three to five clinicians, an EHR with open connections. Compare a suite against a scribe plus phone tool. If the suite does not push notes into your EHR, or does it badly, choose separate tools with a tested integration.
• Six or more clinicians, or an EHR with its own AI. Start with what the EHR offers, then buy separate tools only for functions it does not do well. At this size the vendors will quote custom team plans, and the contract terms matter more than the list price.
• A practice with heavy phone demand. Judge the phone function first and separately. A busy practice loses most to missed calls, and a scribe saving two minutes a note will not make up for it.
• A practice that dictates and writes letters more than it talks. Look at dictation and letter features before ambient notes; the dictation versus ambient comparison explains which suits you.

The wider question of building, buying or hiring is covered in AI solutions for medical practices: build, buy or hire help; this tutorial assumes you have decided to buy.

One BAA per vendor, and what to ask in each

Every vendor that handles patient information needs its own business associate agreement (BAA), and so does any subprocessor it uses. A suite hides several functions behind one vendor, but the functions may be built on other companies' services: a voice provider for calls, a model provider for notes. Ask for the list.

1. Will you sign a BAA before any real data is shared?
2. Which subprocessors handle audio, transcripts or call recordings?
3. Is our data used to train models, yours or theirs?
4. How long are recordings, transcripts and logs kept, and can we delete them?
5. What is the notice period and process if there is a breach?
6. Can we export everything on request, in a usable format?

With separate tools you will ask these six questions two or three times and may get inconsistent answers. That is not a reason to avoid separate tools, but you should plan for the extra review time. The scribe security tutorial goes through the fuller list, and the questions before buying any AI tool covers the non-security side.

The integration test: one practice day, scripted

Before you sign anything longer than a month, run a scripted day with test patients only. The aim is to find the breaks while changing vendor is still cheap.

1. A test patient phones after hours. Does the assistant answer, capture the right details and book, or hand off, as your rules say?
2. The booking appears in your calendar or EHR within a minute, with the correct appointment type.
3. The test patient arrives. A clinician records a consultation with the scribe and signs the note.
4. The note arrives in the EHR in the correct chart, in the right template, with no copy and paste.
5. A referral letter is drafted and reviewed. Did the scribe or the suite produce it in the format you use?
6. A cancellation text arrives. Does it update the calendar and the waitlist?
7. An administrator removes a leaver and checks that their access and stored data go.

Score each step pass or fail. A suite that fails step 4 is a scribe problem; a separate stack that fails step 2 is a connection problem. The nature of the failure tells you whether the next vendor will fix it. For the phone half, a worked call flow and script set gives you realistic scenarios to feed it, and what an AI answering service costs shows what to expect from pricing.

Example: a suite that passed on paper and failed on Tuesday

Illustration: a three-clinician practice chose a suite for its single price. In the integration test, notes reached the EHR, but appointment types from AI-booked calls arrived as "general", which meant the front desk had to fix 20 a week. The fix was a mapping the vendor supplied after a week of back-and-forth. The lesson is that integration breaks are usually small and repetitive, so test the exact appointment types you use.

Three EHR situations and what they point to

The EHR is the fixed point. Everything else has to connect to it, so decide which of these three situations you are in before you compare vendors.

The EHR has its own AI tools

Ask what it charges, what it does, and how it handles audio and call data. A scribe built into the EHR can drop notes into the right place without any connection to manage. Enterprise offerings of this kind, Dragon Copilot among them, are built for larger organisations, so a small practice should expect the EHR vendor's own add-on instead, if it has one. Compare it to an independent scribe on quality and cost, and buy the phone tool separately unless the EHR vendor offers one you have tested.

The EHR has open connections

This is the easiest case for separate tools. Check the vendor's connection list for each tool you are considering, and ask who supports a broken connection: the EHR vendor, the tool vendor or you. Freed lists an EHR push feature on its Premier plan, and Heidi lists EHR integration on its team plans, so read what is included at each tier rather than assuming it is there.

The EHR is closed or old

Here a suite is attractive only if it can fill the gap, and many cannot. When there is no connection, notes travel by copy and paste, which costs about 20 to 30 seconds per note. At 360 notes per clinician per month that is two to three hours each, a cost to count against any tool. A suite that includes its own scheduling and records may be worth a longer look, but it then becomes an EHR decision, not an AI one, and needs a longer, separate process.

What demonstrations tend to hide

• Clean audio and scripted patients. Ask to hear a call recorded on a poor line and to see a scribe note from a noisy room.
• Integrations shown on a sandbox. Ask to see the connection working with your EHR, even if only in a trial account.
• Averages. A headline such as "saves 70 percent of documentation time" is a claim about a sample. Ask for the sample and the audience.
• What happens on a bad day. Ask what the phone assistant does when it does not understand a caller, and what the clinician sees when the scribe fails to capture.
• Plan boundaries. A cheap plan with a note cap, such as Freed's 40 a month on Starter, may suit one part-time clinician and no one else.

A fallback plan for each function

A tool that fails with no fallback becomes a patient-safety and reputation problem. Write one line for each function before go-live, and test it once.

The last row is where a suite can hurt. If one vendor outage takes down calls, notes and inbox together, you need all three fallbacks working at once. A separate stack lets you lose one function at a time. That is not a reason to avoid a suite, but it is a reason to test the combined fallback.

A ninety-day purchase path for a four-provider practice

1. Days 1 to 10: list the jobs (notes, calls, letters, inbox) and rank them by pain. Decide which one function you will buy first.
2. Days 11 to 25: shortlist two suites and one separate stack. Send each vendor the same written questions, including the BAA, and wait for answers.
3. Days 26 to 40: run the integration test with the two best candidates, on monthly terms where offered.
4. Days 41 to 70: live use in two clinicians and one front-desk shift. Record minutes saved, errors found, missed calls and any patient comment.
5. Days 71 to 90: decide. Roll out to the rest, add the second function, or drop the option that failed. Write the reason down.

The path costs about 20 to 30 hours of staff time across the practice and, on monthly terms, one to two months of subscriptions for the trial. If a suite wins, you pay a year at a time only after day 90.

A comparison email to send to each shortlisted vendor

We are a four-clinician practice. We want to compare [vendor] with
a separate scribe and phone tool. Please confirm, in writing:
1. Price per provider per month at our size, and what counts as a provider.
2. Whether notes are pushed into [our EHR] and what that connection costs.
3. Your BAA, subprocessors, retention and training policy.
4. What the phone assistant does when it cannot answer.
5. How we export our data and what happens when we leave.
6. Terms for a one-month paid trial.

Compare replies side by side. Look at what is missing, not only what is said: an answer that skips question 3 usually matters more than a low price.

Weakest-link scoring before you accept a bundle price

Score each function out of five, on its own, using the results of your test. A filled-in example:

Here the suite is strong where the practice has a gap (inbox, one agreement) and weak where it cares most (scribe quality). A sensible answer might be a suite for phone and inbox, with a specialist scribe. Mixed stacks are legitimate, so long as you count the agreements and the connections they bring.

Lock-in, contracts and the first year

The cost of leaving is hard to see on day one, so ask for it now. Check the following in writing:

• Monthly terms for the first three to six months, even at a higher price, until the integration test has run in real use.
• Export of notes, templates, call recordings and logs in a readable format.
• Price protection: whether per-provider or per-call prices can change mid-term.
• What counts as a provider. Locums, part-timers and trainees can change the bill.
• Deletion of stored data within a stated period after you leave.

Annual billing often brings a lower price, as Heidi's $110 and DeepCura's $999 a year do, but it also commits you for a year. Take it only after the integration test and a month of live use.

When to start with only one function

Many practices should not buy a suite or a stack on day one. If the pain is notes, begin with a scribe on its own, choose it for scribe quality, and see what the vendor adds later. If the pain is missed calls, begin with the phone side. Add the second function only when the first has a measured result. The cleanest way to avoid regret is a small first purchase that can be expanded or abandoned at low cost. If you would like help scoring the options against your own EHR and calls, that is what my AI implementation consultation covers.

Questions practices ask when comparing bundles

Is an all-in-one AI suite cheaper than separate tools?

Often on the invoice, not always overall. DeepCura lists its scribe, receptionist, fax and scheduling bundle at $129 a month per provider seat, less from the third seat, while a scribe at $79 to $110 plus a phone service priced by call volume usually costs more at list price. Check what the suite's monthly credits cover, and count integration time and switching cost too.

Do I need a separate BAA for each tool?

Yes. Every vendor that touches patient information needs its own agreement, and any subprocessors it uses should be named. One suite means one BAA to review, which is a real advantage; separate tools mean several, each needing the same checks.

What if my EHR already has an AI scribe?

Compare its cost and fit first. A scribe built into the EHR avoids copy and paste and extra logins. If its price or quality is poor, an independent scribe that pushes notes into the EHR may still be better, but test that integration.

How do I avoid lock-in?

Ask whether notes, templates, call recordings and logs can be exported, in what format, and what happens to stored data when you cancel. Prefer monthly terms for the first year and annual terms only after the integration test passes.

Should the phone assistant and the scribe come from the same vendor?

Only if both meet your standard. A suite that is strong at notes and weak at calls leaves you with a weak link. Score each function on its own before you accept the bundle price.

Further reads

• AI Answering Service for a Doctor's Office: After-Hours Call Rules — After-hours call rules for the phone half of the decision.
• AI Chatbot for a Doctor's Website: Scope, Disclosure, Handover — Whether a website chatbot belongs in the bundle.
• Best AI Receptionist for a Medical Practice: What Each Type Does — The kinds of receptionist tool and what each does.
• AI Scribes for Doctors: Six Note Errors to Catch Before You Sign — Note errors to test for in any scribe you shortlist.
• Cliniko for Doctors: Fit, Cost and the AI Tools That Sit Around It — How an EHR-style practice system and AI tools fit together.
• How a Small Clinic Can Use AI at the Front Desk Safely — Safe front-desk uses of AI in a small clinic.
• AI Medical Coding Software: Audit Trails and Upcoding Risk — Suggest or submit? Who signs, what the audit trail must record, how to spot upcoding drift and what to test on 100 old charts.
• AI Medical Transcription Services vs Human Transcribers for Small Clinics — Per-minute quotes, turnaround, BAA and an error-check step, plus a five-dictation test that shows which service suits your clinic.
• Best AI Scribe for Urgent Care: Fast-Visit Notes and Coding Checks — How to pick an AI scribe for walk-in clinics: five-minute visits, templates per complaint, coding suggestions and a one-week test.
• Best AI Scribe for Veterinarians? A Ten-Consult Test and Scorecard — A ten-consult trial plan and scorecard for vet practices: SOAP accuracy, PIMS export, species terminology, consent wording and price per vet.
• Best AI Software for Therapists in a Group Practice: Seat Costs — Per-seat list prices for SimplePractice, Jane, Carepatron and others, with a five-clinician cost table and the admin controls owners need.
• Best AI Tool for Medical Billing: Scrubbing, Coding, Denials — Pick an AI billing tool by job: claim scrubbing, coding suggestions, denial work and patient statements, with list prices and what to check first.
• Dermatology EHR With AI: What Is Real for a Small Clinic — Dermatology EHRs sorted into real AI (ambient notes, code suggestions) and marketing (diagnosis from photos), with a cost sum for a small clinic.
• AI App for Medical Providers: Standalone, Add-On or Built-In? — Standalone app, add-on or built-in feature? A vendor-neutral look for small medical practices on cost, data flow, switching and fit, with a decision table.
• AI Receptionists for GP Practices: A Four-Week Trial and Stop Rules — A GP practice's four-week AI receptionist trial, week by week: what to measure, what counts as a failure and the stop rules agreed before it starts.
• AI Scheduling for Doctors: Appointment Types and Waitlist Rules — How to configure AI scheduling for a doctor's practice: appointment types, buffers, overbooking limits, waitlist fill rules and a safe link to the EHR calendar.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: DeepCura, Freed, Heidi and Smith.ai product and pricing pages, plus Microsoft's Dragon Copilot page, checked 4 November 2026; vendor claims quoted as claims.