In 2017, Covenant Health began exploring a peculiar possibility. What if the insurer could see the patient’s story while the hospital was still writing it? The Knoxville health system was using Xsolis’s CORTEX software, now called Dragonfly. It wanted less paper, less chasing and a better conversation about medical necessity. In 2018, Humana joined it in a shared review process. Nurses on both sides could work from the same clinical picture.
- The job: help hospitals and insurers assess medical necessity and coordinate patient-status decisions.
- The mechanism: continuously updated clinical data, predictive scores and agreed rules for automating straightforward cases.
- The economics: enterprise software, optional capabilities and clinical services. Savings depend on how the workflow changes.
Usually, a hospital bill arrives after the care. Here was an attempt to move the argument earlier, when the information was fresh and the people involved could still act on it. Bureaucracy had acquired something approaching good manners: both parties were invited to look at the same thing.
The insurer comes into the room
Xsolis occupies a consequential corner of healthcare technology: utilization management, where teams assess whether a patient’s care setting and services are appropriate. An inpatient designation and an observation designation may describe care delivered inside the same hospital. They carry different administrative and reimbursement consequences. Getting the designation right requires clinical information, documentation and judgment.
The traditional workflow gives this work plenty of opportunities to become repetitive. A hospital team reviews a chart. A payer team reviews submitted information. Someone needs a missing detail. Someone makes a call. A clinical professional spends another part of the day searching rather than deciding.
Xsolis connects electronic medical record data to a shared review environment. Its proposition is specific: give both parties a current clinical picture, organize the evidence and concentrate human attention on cases that need it. The company’s hospital users include utilization-review nurses, case managers, physician advisors and revenue-cycle leaders. Health-plan reviewers use the other side of the arrangement.
“We’ve been able to streamline our workflow, save time in our process and improve the ability to communicate medical necessity.”Sherri Ernst, Covenant Health · XCHANGE 2019
The 2018 payer relationship matters because it changed the work the software could do. Hospital analytics could help one side prepare its case. Shared analytics could help both sides settle it. That is a different commercial proposition, requiring cooperation as well as computation.
A number with a job to do
At the center is Xsolis’s Care Level Score, or CLS. The model uses clinical information to predict medical necessity and the likely appropriate patient status. As an encounter evolves, the picture updates. Staff can prioritize reviews rather than treat every chart as an equally urgent mystery.
The score also gives payer and provider teams a reference for agreeing thresholds. Under Precision Utilization Management, they can automate determinations for qualifying cases where their historical decisions align. Difficult cases still require attention. The ingenuity lies in identifying where another round of review adds little.
- 01Clinical recordRelevant data flows from the EMR.
- 02Patient picturePredictions help prioritize the case.
- 03Shared reviewHospital and health plan assess the evidence.
- 04Agreed routeQualifying cases automate; complex cases escalate.
This is where Xsolis differs from a conventional screening-tool purchase. MCG and InterQual are familiar names in medical-necessity review. Xsolis’s pitch combines continuously refreshed predictions, workflow and a payer-provider connection. Buyers must compare the actual jobs each approach performs, including how evidence reaches the insurer and how exceptions are handled.
A prediction still has limits. A Baylor Scott & White study reported 86% correct classification at a defined CLS threshold; performance changed with the cutoff. That is a useful reminder that a score is a decision aid with operating choices. It does not abolish uncertainty, and one study’s result is not a promise for every hospital.
The price of a shorter argument
This is institutional software bought through institutional contracts. Public records offer a rare look at the bill. In April 2023, UMC Health System’s board minutes reported existing Xsolis spending of $516,000 annually. The renewal discussion described additional reporting, analytics and physician-advisor modules, and said a 2022 analysis had found positive return on investment.
UMC’s existing annual spend, reported April 2023. A historical deployment figure, not a current quote.
The University of Alabama System’s 2025 board budget materials describe a four-year Xsolis implementation project with an $8.5 million total project cost and $1.2 million fiscal-year cost. The entry covers UAB Hospital and UAB St. Vincent’s, describes volume-based pricing and says increased usage raises recurring costs. Project budgets and subscription bills are different measures; neither should become an imaginary universal price tag.
The buying logic is nevertheless legible. A hospital pays to reduce review effort, improve status accuracy and protect appropriate reimbursement. A health plan seeks faster concurrent reviews and less administrative friction. Evaluating that bargain means counting both the software cost and the work that changes around it.
The software comes with people
Dragonfly divides the work into recognizable products. Utilize supports hospital review and prioritization. Align supports payer collaboration. Advise organizes physician-advisor cases. Revenue Integrity Insights connects clinical evidence with denial and payment trends. Navigate, launched in 2025, extends the platform into discharge planning and capacity management.
Xsolis also supplies physician advisory services, either supplementing an internal team or providing outsourced coverage, plus denial services. This hybrid business has a practical advantage: the customer can purchase expertise alongside the system that organizes the case. The expertise includes second-level review, peer-to-peer discussions and compliance support.
Its physician advisory services won Best in KLAS in 2021, 2023, 2024, 2025 and 2026. Naming the years is more useful than wrapping them in a vague claim about a winning streak. The distinction belongs to the services category, rather than proving every algorithmic claim.
Chief executive and co-founder Joan Butters brings healthcare operations and technology experience from Optum and AIM Healthcare. The customer support model includes education, reporting and business reviews. There is a sensible theory underneath that attention: a tool’s value depends on whether people use it fully and whether the process actually changes.
Ten minutes returned to a nurse
Generative AI adds a different kind of assistance. Instead of only predicting a status, it can draft a clinical summary for a reviewer to verify. Beacon Health System tested Xsolis’s initial medical-necessity review tool in a three-month pilot. Results announced in June 2025 put average review time at 4.7 minutes, compared with 15 minutes before - a reported 68% reduction.
17.2 hours of review time at that volume.
The distinction between predictive and generative tools matters. A score helps decide where to look. A summary helps assemble what the reviewer needs to read. Nurses still check, refine and finalize the output. The clinician’s work changes shape rather than disappearing.
On September 1, 2026, Xsolis announced general availability of another capability: a peer-to-peer clinical synopsis inside Advise. It draws out key facts and a day-by-day account before physician advisors discuss denied cases with payer medical directors. These conversations often have tight deadlines. Reconstructing a long chart is an expensive way to spend the preparation window.
Shared data has a second obligation
In June 2026, Xsolis disclosed a security incident resulting from a January 20 targeted phishing attack, detected January 22. An unauthorized actor acquired files. Depending on the person, exposed information could include names, insurance information, Social Security numbers and medical treatment information. An Indiana regulator’s register lists 1,396,519 people affected.
The company said it contained the access, investigated with outside experts, notified law enforcement and added security measures. As of its June 5 notice, it reported no known actual or attempted misuse. The incident belongs in any assessment of the business: a shared clinical record is useful precisely because the information is sensitive and consequential. Protecting it is part of the product’s value.
Start with the cases nobody disputes
The lesson other organizations can copy is concrete. Bring the counterpart into the process. Agree what makes an uncomplicated case uncomplicated. Measure the existing review burden. Automate within that agreement, and preserve skilled attention for the exceptions.
It needs conditions to hold. Data must arrive reliably. Payers and providers must participate. Reviewers must trust the evidence enough to use it, while retaining judgment. Discharge predictions need teams capable of addressing actual barriers; a forecast cannot conjure an available placement or transport.
Xsolis reported more than 615 hospital and health-system customers in September 2026. Its most interesting idea remains smaller than that number: before building machinery to argue faster, examine how much of the argument people already agree to skip.