Pretty Good AI vs XCaliber Health

Two companies making the same argument about point solutions

XCaliber Health argues that practices have spent twenty years accumulating EHRs, billing platforms and scheduling systems without ever acquiring the operating system to coordinate them.

We make a version of that argument constantly. A practice running a separate vendor for reminders, one for reputation, one for intake and one for the phone has four contracts, four integrations and four places for a workflow to break.

So this is not a comparison between a consolidator and a collection of point tools. It is a disagreement between two consolidators about what should be consolidated.

The disagreement, stated plainly

XCaliber consolidates horizontally. A data gateway normalizes information out of every system an organization owns, a FHIR resource catalog covers integrations across EHRs, and agents run on top of that layer. The unit of consolidation is the organization.

We consolidate vertically. One system, athenaOne, integrated as deeply as the API surface allows, with every front-office workflow running on that single integration. The unit of consolidation is the practice’s actual operating system.

Both are legitimate. They suit different organizations, and the choice is usually decided by a question that has nothing to do with either vendor: how many systems does your work genuinely live in?

What the horizontal approach is built for

If you are a health system with several EHRs from acquisitions, ancillary systems that do not talk to each other, and data that has to be assembled before anything can act on it, then the integration is the problem. A gateway and a normalization layer are the architecture for that problem.

The same is true if you want agents covering clinical workflows alongside operational ones. XCaliber publishes agents across the care continuum with a human-in-the-loop structure and stated levels of autonomy per agent function. Clinical decisions stay with your care team. Everything around them, from the first call through the visit and the bill, is in scope.

What to verify about the horizontal approach

FHIR is what makes multi-EHR coverage possible. It is a standard for exchanging health data, and normalizing to it is the reason one platform can work with many systems.

How deep that coverage reaches into each system’s configuration is the part worth understanding before signing anything.

A large share of what an athenaOne practice actually runs on is not standardized clinical data. Appointment types. Scheduling templates and the slots inside them. Provider groups. Departments. The order and tickler queue. Document classes. Those are athenahealth-specific configuration, and they are exactly the objects a front-office workflow has to touch to finish rather than start.

We hold production access to 820+ athenaOne APIs, and that number is not a boast about engineering effort. It is the reason an agent can read the live slot table, book the right appointment type with the right provider group at the right department, and close the tickler behind it, instead of writing a task telling a human to do those four things.

What an agent working through a normalized layer can reach is a question for XCaliber to demonstrate. Ask them to book your hardest appointment type end to end in athenaOne, including the tickler behind it, rather than describing what the FHIR surface covers.

The complication that exposes the difference

The following is an illustrative workflow, not a customer story. A federally qualified health center has a recall problem.

Orders are generating follow-up requirements. Patients need calling and booking against them. The queue gets worked when staff have time, which means it is not worked.

The rules are not complicated, they are just specific. Certain visit types have to go to a provider in the same provider group as the ordering provider. Some have to be at the patient’s home department rather than whichever site has a slot. Patients with an interpreter requirement need appointment types that allow longer visits, already configured in athenaOne, with no interest in maintaining them a second time in a vendor’s rules engine.

In our product, the automation reads the order queue, applies the configuration that already exists, calls the patients and books them. It works because every object involved is reachable natively.

Two paths still go to a person, and always will. If no compliant slot exists inside the required window, a human decides what to do, because that is a judgment about the patient’s care. And if the ordering provider has left, the reassignment goes to the practice manager rather than being inferred.

That workflow is not exotic. It is the ordinary shape of front-office work, and it is why the ceiling on integration depth matters more than the breadth of systems covered.

Reading their marketplace record

XCaliber has 138 athenahealth Marketplace practice connections against our 45, which is a real lead and worth noting rather than explaining away. Their listing dates from May 2025, ours from September 2025.

Two things temper it. They have no published marketplace reviews and we have 22 ratings, so there is nothing to read about how those connections are going. And their athenaOne connections concentrate heavily in family practice, multi-specialty, internal medicine and FQHC, so a specialty practice should ask for a reference in its own shape.

Neither number settles anything. Ask both vendors to complete your hardest workflow end to end in an athenaOne environment and watch precisely where it stops and what it hands to a person. That single test tells you more than either marketplace page.

The honest summary

If your systems are plural and the integration itself is the problem, XCaliber is built for that and we are not.

If athenaOne is where your practice runs, the consolidation you need is not across systems, it is across the vendors sitting on top of the one system you already have. That is what an athenaOne-only platform on 820+ athenaOne APIs is for, and the pricing is published, month-to-month, with the first 30 days live free.