Pretty Good AI vs Health Note

The closest comparison on the board, and the one worth reading slowly

Most vendors we get compared with are doing something adjacent. A reputation platform, a check-in kiosk company, a scheduling incumbent that added a phone agent.

Health Note is not adjacent. It answers the call, books the visit and writes the result into athenaOne, which is the same sentence we use about ourselves. When two products describe themselves identically, the comparison is decided by scope rather than by feature checklists, and scope is where these two diverge.

Two differences are worth putting side by side before anything else.

Where the two products differ

Both products cover voice, text and web chat. We put the engineering into a single athenaOne integration, and into what happens after the conversation ends.

Our voice agent handles the call and completes the work in the chart before the caller hangs up. Secure two-way texting and web chat for patients who prefer to type are live on the same athenaOne integration. If your board has already decided that patients text and the front desk needs to text back, we can show it running in production.

They also write clinical documentation. Clinical decisions stay with your care team. Everything around them, from the first call through the visit and the bill, is in scope, including booking the appointment, routing the request, chasing the form and working the order queue.

Saying both of these plainly is more useful than a comparison table that quietly omits them.

Where an athenaOne-only integration goes further

Health Note works across several EHRs. Its athenahealth Marketplace listing describes athenaOne workflows in detail, and its own homepage leads with case studies on other systems.

That is a reasonable business to build. It also has a structural cost, and the cost is depth. A vendor supporting several EHRs builds to the workflows those systems have in common. We hold production access to 820+ athenaOne APIs, which is what it takes to complete work rather than start it.

The difference does not show up on a simple booking. Both products handle that well.

It shows up on everything downstream of the booking. Closing the tickler. Reading the order queue and calling the patients on it. Filing the referral against the right document class. Prepping the eligibility check and the prior authorization. Working denials and appeals. Those workflows are expressed in athenaOne’s own concepts, and an integration built for several systems tends to reach the point where it creates a task for a human at exactly the place the work was already stuck.

The complication that decides these deployments

The following is an illustrative workflow, not a customer story. A multi-site pain management group has a recall problem that looks like a phone problem and is not.

Its orders queue holds procedure follow-ups that have to be scheduled inside a window tied to the previous injection date, at the same location, with the provider who performed it unless that provider has left. Staff work the queue by hand when they have time, which in practice means the queue grows.

The automation that works never touches the clinical decision about whether the follow-up is appropriate. That is already made and recorded as an order. What it does is read the order queue, apply the scheduling rules that already exist in athenaOne as appointment types and provider group settings, call the patients on the list, and book them.

Two cases go to a person every time. If the window has already closed, a human decides what happens next, because that is a judgment about the patient’s care and not the automation’s call. And if the performing provider has left the group, the reassignment goes to the practice manager rather than being guessed at.

The reason that workflow can run at all is that every object it touches, the order, the appointment type, the provider group, the location rule, lives in athenaOne and is reachable. That is the argument for depth over breadth, and it is the whole argument.

Reading the marketplace record honestly

Health Note has 45 practice connections and 51 reviews at a 5.0 average as of 7 September 2026. We have 45 connections and 22 ratings, all five-star, as of October 2026. Their listing dates to March 2019 and we have been an athenahealth Marketplace partner since 2025.

The connection counts are close. The review gap is mostly six years.

The better use of that page is to read the reviews rather than count them. Look for a practice in your specialty at your size describing the workflow you are actually trying to fix. Their athenaOne connections concentrate in multi-specialty, family practice, orthopedic surgery, FQHC and pain management. If you are in one of those, there is probably a relevant review to read, and you should read it.

We would rather you evaluated us on that basis than on a total.

What to ask both vendors

Take your hardest appointment type to both. Not the easy new-patient visit, the one requiring a specific provider at a specific location with a preceding visit. Ask each vendor to book it in an athenaOne environment rather than a sandbox.

Then ask the question that separates the products: what happens when the agent cannot finish. Ask to see the handoff, who it reaches and what the person receiving it sees. Most front-office automation is judged on the calls it completes and breaks on the calls it does not.

Ask both of us for current audit reports too, and read the scope section on each. Ours: The Pretty Good AI platform has attained HITRUST i1 certification. Pretty Good AI signs a HIPAA Business Associate Agreement (BAA) before any patient data is touched. SOC 2 Type II and ISO 27001 audit reports are available upon request. GDPR-aligned handling.

Designed in 2019, versus designed for agents

Health Note has been building since 2019, before generative models could hold a conversation. Its intake product is a structured questionnaire engine, and it is a good one, because a structured questionnaire was the best available way to collect that information at the time.

Our starting point was different. The agent talks to the patient, works out what is actually being asked, and writes the result into athenaOne. A platform whose core abstraction is a form can add an AI layer over the form; it cannot easily stop being a form.

The honest summary

Clinical documentation is not something we build; that stays with your care team and the tools they already use.

If you are standardized on athenaOne and the work you need finished lives deep in the chart, after the call rather than during it, that is what an athenaOne-only platform on 820+ athenaOne APIs is for. Pricing is published, it is month-to-month, and the first 30 days live are free, so the comparison does not have to be settled in a slide deck.