How to Tell Whether an athenaOne Add-On Can Do the Real Work

Every athenaOne add-on in the market says it integrates with athenahealth. The phrase is doing enormous work and almost none of it is specified. It can mean a vendor reads open appointment slots and writes a booking, which is a genuinely useful but narrow thing, or it can mean the vendor operates inside your work queues the way a staff member does. Those are different products at the same price point.

Practices have learned this the expensive way. The pattern is familiar enough that administrators describe it before you ask: three or four point tools signed over two years, each of which demoed beautifully, each of which handles a slice and hands the difficult remainder back to the same overloaded person.

The remainder is where the hours are. A tool that books an appointment but cannot open a case, assign it to the right queue, attach what it learned on the call, and close it when the work is done has not removed a workflow. It has inserted itself into the middle of one and left the ends for staff.

So the practice ends up doing the original job plus the new job of watching the tool. That is the honest description of what most integration claims deliver, and it is why the useful evaluation question is not whether a vendor connects to athenahealth. It is how much of athenaOne it can actually operate.

Ask for the ratio, not the number

The first question to ask any vendor is how many of athenahealth’s APIs they support, and the answer only means something as a fraction.

athenahealth exposes roughly 800 endpoints across scheduling, patient and chart data, orders and tasks, documents, billing, and communications. Pretty Good AI supports 440 or more of them. A vendor supporting twelve is not a smaller version of that; it is doing a fundamentally different thing, because the twelve will be the calendar endpoints and nothing behind them.

The denominator is what makes the number interpretable. Any vendor can say they support dozens of APIs. Ask what share of the platform that represents and which categories are missing, and the conversation gets concrete fast.

The reason this matters operationally rather than technically: breadth is what determines whether a workflow can finish. Booking lives in one category, the case that tracks the follow-up lives in another, the document that has to be attached lives in a third, and the eligibility check lives in a fourth. A vendor entitled to one of those four can only ever hand you the other three.

Six questions that separate reading from operating

These are worth asking verbatim, and the quality of the answers is more informative than any demo.

Can you open a patient case in athenaOne, and can you assign it to a specific work queue? Can you close one, with a reason that shows up in reporting? Can you see what is sitting in a provider’s inbox and route around a backlog? Can you check eligibility before you confirm a slot rather than after? Can you attach what happened on the call to the case, so the staff member who picks it up is not starting over? And when your automation cannot finish something, where exactly does it land, and who owns it from there?

The last question is the one vendors handle worst and the one that predicts your experience most accurately. Everything gets handed off eventually. A vendor that has thought hard about the handoff will describe the queue, the attachments, and the reason code without hesitating. A vendor that has not will describe an alert.

There is a related question worth asking about routing specifically. The primary provider field in the chart is stale almost everywhere and cannot be trusted to route a call. Ask what the vendor falls back on. The correct answer involves who has actually seen the patient recently, not the field.

The sector’s own experience of buying AI is instructive

It helps to know that the disappointment is common rather than personal.

An August 5, 2025, MGMA Stat poll of 244 applicable responses found 71% of practice leaders reported some use of AI in patient visits, but among those using it, 44% said it had not reduced staff workload, 39% said it had, and 17% were unsure. Adoption is nearly universal and the workload result is close to a coin flip.

That spread is not mysterious. A tool that automates a step inside a process leaves the process, and the person, exactly where they were. A tool that owns a workflow from intake to closure removes hours. Both of them are marketed as AI, and both integrate with athenahealth.

The evaluation question that separates them is the depth one. What can this thing finish without a person, and when it cannot finish, does it hand off cleanly or does it just stop?

Point the evaluation at where your hours actually go

Practices tend to evaluate against the workflow that is most visible rather than the one that is most expensive.

A March 10, 2026, MGMA Stat poll of 294 applicable responses asking practice leaders which phone tasks consume the most staff time put eligibility and prior authorization at 45%, ahead of scheduling at 31%, intake at 9%, and prescription refills at 6%. For an internal medicine practice with a large established panel, the verification and refill-request traffic is relentless in a way that never shows up in a demo built around booking a new patient.

So run the demo against your own queue. Ask the vendor to handle an eligibility check on a real appointment type, open the case, route it, and close it. Ask what happens with a refill request that arrives at 11pm, and specifically where it lands for the clinical staff who will act on it in the morning.

A vendor whose product only performs on the scheduling third will still perform on the scheduling third. You will simply have bought a third of a solution and still have the 45% on the phone.

What depth buys you on an ordinary Tuesday

The concrete version of this is worth spelling out, because depth sounds abstract until it is a specific afternoon.

A patient calls asking for a doctor by name. That provider is not accepting new patients, which is a conversation problem before it is a data problem: the caller needs language that redirects them to a provider with availability without sounding like a brush-off. The system needs to know who is open, in which department, for which appointment type, then book it.

Behind that single call sits the scheduling surface, the provider group, the appointment type catalog, an eligibility check, and a case opened so the front desk can see what was offered and declined. A shallow tool does the booking and leaves the rest. A deep one does the whole sequence and stops at the one place a person is genuinely needed.

That point exists in every workflow and it is not a weakness. Anything requiring judgment about the patient’s care goes to a clinician, immediately, with the context attached. The measure of a good add-on is that the handoff is deliberate and complete rather than the place where the automation quietly gave up.

Key Takeaways

Ask for API support as a ratio of athenahealth’s roughly 800 endpoints, because the denominator is what makes the number interpretable.
Test whether a vendor can open, assign, and close a patient case with a reason code, not just read a calendar.
Make every vendor describe exactly where an unfinished item lands and who owns it, since the handoff predicts your experience better than the demo.
Ask what the routing falls back on when the chart’s primary provider field is stale, which it is almost everywhere.
Run the demo against eligibility and refill traffic rather than new-patient booking, because that is where the phone hours actually go.
Treat a tool that automates one step inside a process as a partial answer, and count the remainder as work your staff still owns.

The difference between an athenaOne add-on that reads your calendar and one that operates your practice is not visible in a demo, and it is not visible in the phrase integrates with athenahealth. It shows up in the ratio of the platform a vendor can reach, in whether a workflow can finish, and in how carefully the handoff to a person was designed. Ask the six questions before you sign.

Sources