We sat down with Chief AI Officer Spencer Allee and Chief Technology Officer Curtis McGinnis for an honest, plain-language discussion on how AI and automation are actually working in healthcare revenue cycle and what health systems and hospitals should understand before evaluating any technology in this space.
Q: There is a lot of AI noise in revenue cycle right now. How do you think about separating what is real from what is marketing?
Spencer Allee: The skepticism is completely warranted. AI has become a catch-all term that gets applied to everything from genuinely sophisticated machine learning and large language models (LLMs) to a slightly smarter “if-then” statement. What I look for, and what I would encourage any health system leader to look for, is specificity. What is the model actually doing? What data was it trained on? How does it perform when the documentation is messy, inconsistent, or non-standard? Those are not unfair questions. They are the baseline. Any AI that is production-ready in a clinical and financial context should have clear, confident answers to all of them. If it does not, the technology is probably not as mature as the marketing suggests.
Q: Walk us through how a claim moves through Aspirion’s platforms from intake to resolution.
Curtis McGinnis: At its core, what we built is an end-to-end operational system that makes sure the right work gets to the right person at exactly the right moment, and that nothing falls through the cracks because the volume got too high or a deadline snuck up on the team. Compass is the engine driving all of that orchestration. When a claim comes in, it handles intake, reads the denial letters and correspondence, and routes the account based on what it actually needs, whether that is a clinician, an attorney, or a payer negotiator.
It is also prioritizing by recovery potential and managing filing deadlines in the background. The reason that matters is that the failure point in most recovery operations is not the work itself. It is the operational infrastructure that should have gotten the claim to the right person in the first place. That is the problem Compass was designed to solve.
Q: DocIQ processes more than 50,000 documents a month. What is AI actually doing, and why does accuracy at that scale matter so much?
Spencer Allee: Medical records are one of the hardest document types for AI to navigate well. They are not clean or structured. You have conflicting clinical notes, non-standard tables, imaging reports written in shorthand, and narrative documentation that buries the relevant fact deep in a lengthy record. Building AI that can reliably extract clinical evidence from that kind of complexity and cite every finding back to its exact location in the source document requires purpose-built architecture trained specifically on healthcare data. That is what DocIQ is.
DocIQ was not adapted from a general-purpose model. It was engineered for this problem from the ground up, trained on proprietary claims data across hundreds of payers and providers. The 99.7% extraction accuracy is a direct result of that specificity. And that accuracy matters enormously in practice because an appeal is a legal and clinical argument. One misread diagnosis or missed finding gives a payer grounds to dismiss an otherwise winnable case. At 50,000-plus documents a month, near-perfect precision is not a nice-to-have. It is what makes the model safe to rely on at that scale.
Q: What does processing that volume actually mean for a health system in practical terms?
Curtis McGinnis: It changes the economics of what is worth pursuing. Right now, most health systems are making an informal decision every single day that certain accounts are not worth working, not because they could not be won, but because there are not enough hours to get to them. Lower-dollar, high-volume claims age out and get written off. That is not a strategy, it is just the math of manual capacity.
When Compass is routing efficiently and DocIQ is absorbing the document review work at scale, that ceiling lifts. Accounts that were previously off the table become viable. We are not just making teams faster at the work they were already doing. We are expanding the pool of claims that can realistically be pursued. That is a fundamentally different conversation than efficiency.
Q: Underpayments are often described as the hidden revenue problem. How does ContractIQ surface what health systems are missing?
Spencer Allee: Underpayments are invisible in a way that denials are not. A denial shows up in a worklist. An underpayment looks like a paid claim. The account closes and nobody asks whether the payment was actually correct under the terms of the contract. Answering that question systematically requires parsing thousands of pages of payer agreements against individual claim data and doing it fast enough to dispute the variance before the window closes. That is a genuinely machine-scale problem. ContractIQ was built to solve it.
ContractIQ converts dense, complex payer contracts into structured, queryable logic, calculates expected reimbursement based on actual contract terms, and flags the discrepancy with a citation to the exact clause that supports the dispute, in seconds. The health systems missing these underpayments are not missing them because they are not paying attention. The complexity is just beyond what any manual process can handle consistently and at volume.
Q: Why did Aspirion build these platforms in-house rather than using third-party tools?
Spencer Allee: From an AI perspective, the answer is straightforward. The problem is too domain-specific for general tools to solve with the accuracy we need. Healthcare documentation has its own complexity, its own language, its own failure modes. A model trained on general text will make mistakes in that environment that are not acceptable when the output is a clinical or legal submission with financial consequences. DocIQ and ContractIQ were trained on proprietary data built specifically for this work. That is not a differentiating feature we added on top. It is the foundation the accuracy is built on.
Curtis McGinnis: From an engineering and architecture standpoint, building in-house means we own how the system evolves. Payer behavior changes constantly. Denial patterns shift. Contract structures get more complex. New prior authorization requirements appear. When that happens, we update our platforms directly. We do not file a support ticket and wait for a third-party vendor’s next release cycle. That responsiveness compounds over time. The platform our clients use today is sharper than it was a year ago because every claim worked, every outcome recorded, and every payer pattern observed goes back into making the system better.
And that kind of continuous improvement is only possible when you own the technology end to end. Which is why we’ve invested heavily in building out our engineering organization at Aspirion—honestly, some of the best engineers in the world, all focused on one thing: helping providers get paid accurately, quickly, and transparently.
Q: There is a lot of conversation in healthcare about AI replacing people. How does Aspirion think about the human role in this model?
Spencer Allee: As a Chief AI Officer, I think about this a lot, and my honest view is that the replacement framing misses the point entirely. The right question is not whether AI can replace human judgment. It is whether you are deploying each where it genuinely excels. AI is extraordinary at volume and precision, processing tens of thousands of documents without fatigue, extracting clinical facts consistently, pricing claims against contract terms across an enormous portfolio. Our team of clinicians, attorneys, and claims specialists brings what AI can’t: judgment. Knowing which clinical argument is most persuasive for a specific payer, recognizing when a case deserves more attention, knowing when to push harder.
That human-in-the-loop model is what allows us to scale in ways that weren’t previously possible. Historically, about 60% of denials were never pursued—not because they weren’t worth it, but because the economics didn’t work. When AI and experts work together, that changes. The AI surfaces what matters, and our people apply their expertise where it counts most. The result is that we can now recover revenue from that long tail of denials that used to get left on the table.
What results are health systems actually seeing?
Curtis McGinnis: The metrics I come back to most are overturn rate and speed because they are the ones that translate most directly into real financial impact. A 20%+ lift in overturn rates means cases that were lost before are now being won, and those are real dollars recovered that would have been written off. Payment arriving 20 days sooner is cash flow that was not there before. And moving 2.2 times faster from placement to first appeal means we are consistently beating deadlines on accounts that used to time out before anyone got to them.
What I always emphasize is that those numbers are a system outcome. They come from Compass, DocIQ, and ContractIQ working in lockstep, automation creating velocity, AI creating evidentiary quality, and specialists ensuring every output holds up when it matters most. Pull any one piece out and the results drop.
Q: What is the one thing you wish more revenue cycle leaders understood going into an AI evaluation?
Spencer Allee: Accuracy, and the specific conditions under which accuracy holds. It is easy to show impressive numbers in a controlled demo environment with clean, well-structured data. The harder question is how the model performs when the documentation is non-standard, the records are incomplete, or the clinical language is ambiguous. That is the real environment. Probe that in every evaluation. Ask to see performance data on messy, real-world documentation. Ask how errors are caught and what happens downstream when one slips through. Those conversations will tell you far more about production readiness than a polished presentation ever will.
Curtis McGinnis: My answer is scope. Most technology conversations in revenue cycle are about doing existing work faster. That is a legitimate goal and real value. But the more transformative question is whether the technology is changing which work can be done at all, whether it is making previously unworkable accounts viable to pursue. That is where the compounding financial impact lives, and it is the conversation I wish more revenue cycle leaders were pushing toward when they sit down to evaluate a new platform.
At Aspirion, we believe in the art of the possible. We refuse to be bound by current limitations, both in terms of programmatic constraints and workflow constraints. The evolution of what we are building is incredibly exciting!
To learn more about how Aspirion’s Compass, DocIQ, and ContractIQ platforms work together, visit www.aspirion.com/technology.
