Spotlight: JJ Martin Quesada @ Foresight RCM
A spotlight is a short-form interview with a leader in health tech. In this spotlight, you'll hear from JJ Martin Quesada, CEO of Foresight RCM.
What does Foresight RCM do?
Foresight automates the revenue cycle end to end for specialty clinics: from benefit verification and discovery to prior authorization (both pharma and medical), claims, medical necessity and reimbursement scrubbing per payer and specialty, reconciliation, and live analytics. We handle most routine tasks by leveraging best-in-class systems tailored to a clinic's specialties and operational reality, bringing in expert human judgment where it's needed.
We're often brought in to handle the most painful workflows where it's hard to scale teams and other automation tools have been tried and have failed. Clinics often ask us to handle more workflows upstream and downstream of the initial engagement.
Prior authorizations delay patient treatment; denied claims sit in a backlog and become write-offs; and routine eligibility checks fail to tell the whole story. A check can come back "inactive" just because the patient got moved to a new plan, or "active" while hiding a behavioral health carve-out.
We layer Stedi on top of a clinic's eligibility stack so when a plain check fails or looks off we can run insurance discovery or an MBI lookup for Medicare patients. This logic often returns coverage that is actually active so we can mark a lapsed policy as inactive and the new one as current with effective dates. Anything short of a clean match goes to a person to review together with what we found.
Engagements start by mapping a clinic's data and workflows to the specifics their payers ask for. Foresight then runs boring but important checks: notes must be signed, clinicians must be contracted with the payer, and payer quirks like custom taxonomies or modifiers must be applied. All automations are gated by simple, easy-to-understand scores: this is right, this is wrong, or not sure. Every "not sure" goes to a human together with the reason (say, the payer wants two failed medication trials with dates and doses, but the chart is missing dates) for final judgment calls.
With the work pile reduced, we give staff the opportunity to look over cases that genuinely need attention while making room for fewer mistakes. Specialists get drastically shorter queues, ranked by priority. Learnings are captured for continuous improvements: e.g., if 26% of prior auths fail due to missing comorbidity data, it might be time to add these questions to the intake flow.
How did you end up working in health tech?
I started working in AI when I first saw its promise 10 years ago and have always wanted to apply it to impactful problems in healthcare. I worked on machine learning and strategy at McKinsey, helped launch consumer products at Google, and built AI solutions in highly regulated industries. I started Foresight when an acquaintance told me about his struggles handling hundreds of thousands of prior auths and payers' byzantine requirements around them. The technology had come far enough at this point while patients were waiting too long to access care. In the US, for every hour a clinician spends with a patient, nearly two more go to desk work, which also contributes to their burnout.
I realized that this was a messy space with the potential to positively impact millions of people, and I really wanted to help. I've always said I'd love to cure cancer, but I'm not the right person to do so. I can, however, help tackle administrative burden, so I moved to New York and started Foresight.
How does your role intersect with revenue cycle management (RCM)?
My job starts by understanding clinics' operational issues and prioritizing what to tackle. We examine denials, documentation requirements unique to their treatments and payers, and look for gaps in their workflows, data, and systems. We then adapt our technology to that reality as I've often seen chaos being caused by generic automation tools that cannot deal with the realities of, say, scattered data or vulnerable patient populations with constantly changing insurance coverage. I also decide where to draw the line between automation and human review. On one side of that line is efficiency and on the other side are false positives and delayed access to care or reimbursement.
The hard part of this problem is not the technology but the data behind it. The evidence each payer demands is scattered across charts; rating scales are buried in notes; medication trial history is a paragraph of prose; some documents are physical scans. Even with perfect data we have to account for the fact that some payers accept the depression scale a clinic already uses, the next one doesn't, and a third wants a baseline score taken within the past four weeks. So before automating anything, we map where each of those facts lives today and what's missing generally against what payers expect.
Payer policies are moving targets: they're published as PDFs that run to hundreds of pages, and they change without warning. We keep a live library of each client's active payer roster, find their policies, and get alerted when one changes before it becomes a new rule in the system. The best denial management strategy is to never get denied in the first place. A lot of problem-solving goes into making sure our model knows what data to ingest and from where, which is our responsibility to handle.
Sometimes what's needed isn't software at all: we've found held-up revenue because a payer contract needed an amendment and simply picked up the phone to call the payer ourselves, handing an amendment package to the clinic that's ready for signature. You have to get your hands dirty in this industry.
What do you think RCM will look like two years from now?
The reviewer on the other side of the clinics is going to be a machine way more often. Medicare is piloting AI-assisted review in six states as we speak. With payers leading the automation run, we see numbers like Medicare Advantage overturning 80.7% of appealed prior authorization denials in 2024, while only 11.5% of denials were ever appealed. Payers have AI denying claims, and providers can't keep up with their billing needs. In two years, whether that gap closes depends largely on whether providers will embrace automation too.
Some EHRs are slowly building their own automation, and some provide connectivity for companies like Foresight to bring the intelligence layer. Other EHRs aren't doing either, and they will soon struggle to retain customers as providers migrate to solutions that don't keep them isolated from the rest of the world.
There will also be more companies like Foresight, tackling the entire cycle rather than offering point solutions. Right now we see a lot of vendors that only handle prior auths, eligibility, or denials. That's already changing as providers get tired of overly confident solutions that don't talk to each other. In RCM, accuracy across steps changes everything. If you chain three point solutions offering 95% accuracy at each step, ~1/7 cases will come out wrong at the end of the process. This is why 63% of organizations say they already use AI in the revenue cycle, but only 15% of them report a positive return so far. The gap is sequencing and customization, not model quality.
One group came to us running five separate vendors: two for benefit checks, an AI phone-caller for the gaps, a prior auth tool, and a clearinghouse, none of which talked to each other. Their staff ended up proofreading confident machines instead of doing their jobs. It's just a matter of time until most clinic owners notice how much custom-tailored end-to-end automation benefits them too.
What I don't think will change is the need for billing specialists. Every deployment I've seen frees up employees who then get immediately absorbed into payer contracting, patient experience, and thorny appeals nobody had the time to address. Two years from now the pile will be smaller, but it's unlikely to fully disappear. We promise robust processes made better by expert human judgment, and that's something that can be delivered today.