How to Automate Xsolis

Xsolis put both sides of the case on one screen

Xsolis makes Dragonfly, a platform for utilisation management: deciding whether a patient in a hospital bed should be there as an inpatient or under observation, and whether the stay continues to be justified as it goes on. It is used by hospitals and health systems, and unusually, by health plans as well.

That last part is what makes it interesting. Historically the two sides looked at the same patient through different windows and argued about what they saw. Dragonfly Align was built for the case where both parties work from the same clinical summary and the same Care Level Score, and reach a determination together rather than by correspondence.

Around it sit the modules for each role: Utilize for the hospital's review team, Advise for physician advisors handling escalated cases and peer-to-peer conversations, and Navigate for discharge readiness and length of stay. The subject in all of them is one patient, one admission and one question that has to be answered while the patient is still in the bed.

A status decision cannot wait for the record to catch up

The reviewer's constraint is not thinking time. It is that the record arrives in pieces, and the decision is due before the last piece does.

A patient comes in from a nursing home, and what was already tried there is in that facility's system, not in yours. A patient had a scan at a freestanding imaging centre last Tuesday, and the report is on the practice's portal. A cardiologist saw them in an office across town and the letter has not been sent. The same patient was admitted at another hospital in the region a month ago, which is exactly the fact that makes this admission look reasonable.

So the nurse works the case with what is in front of them, notes what is missing, and moves on because there are twenty more. The missing piece turns up eventually, often months later, when somebody assembling an appeal finds the document that would have settled the question on day one.

Xsolis

Outside consults on the case Reports from practices and imaging centres that never interfaced, collected and attached before the review rather than during an appeal months later.

Policy changes noticed the week they land Published medical policy pages watched for wording that moved, so a criterion that changed in July is known in July.

The story before the admission The exchange and the referring facility checked for what was already tried, which is often the context that makes this stay make sense.

Dragonfly does the reading a UM nurse has no time for

It is worth being specific about how much of the review this platform genuinely absorbs, because it is the reading, which is the bulk of it.

The Care Level Score assesses medical necessity from clinical data as it lands, and updates rather than sitting still. Generative AI produces clinical summaries so a reviewer opens a case with a readable account instead of a timeline of raw results. The Patient Detail page puts that account in one place. For physician advisors, Advise offers filterable work queues and a generated peer-to-peer synopsis, so an escalated case can be prepared in minutes instead of an afternoon. Bidirectional integration means findings flow back to the EMR rather than living in a second system.

Every one of those capabilities operates on data that reached your electronic record.

The outside cardiologist's letter did not. Neither did the imaging centre's report, the nursing home's medication history, the regional exchange record, or the payer's published medical policy, which is a document on a website that changes without telling anyone.

The consult that settles a case was written in another building

A reviewer is rarely short of judgement. They are short of the two documents that would settle the case, and both exist.

The outside cardiology or pulmonology letter pulled from the practice's own portal and put on the case before the review, rather than found during an appeal. The report from the freestanding imaging centre collected the same way, so absent imaging stops being the reason a stay looks thin on paper.

The regional health information exchange checked for admissions elsewhere in the last few months, so a readmission pattern is visible while it still explains something. The referring nursing home's record read for what was tried before the patient was sent in. Published medical policy pages watched on a schedule, so wording that quietly moved is picked up in the week it moves. Peer-to-peer slots on a plan's own scheduling page taken while the case is still open rather than after it closes.

The safety line here is the one every utilisation department already runs on. Gathering runs unattended and on a schedule. A changed record, a submitted review and anything reaching a patient all wait for a qualified person. Medical necessity is a clinical judgement, made by people licensed to make it.

The missing notes are found before the review starts

Those documents exist. They are just parked on portals the department does not control, and today the fetching falls to a nurse with a worklist, done between the cases that already have everything.

WebRun is an agent that works a real Chrome browser, signed in the way your staff are. It opens the outside practice portal, the imaging centre, the exchange or a plan's published policy pages, reads what is there, and returns it against the patients on today's list.

It runs on your schedule in your own private environment, sessions are not shared between tools, and a workflow can be locked to an explicit list of sites. You can watch a run and stop it at any point.

The workflows below are already built, and each names exactly what it opens.