Latest / Becker’s Healthcare Podcast / How Agentic Revenue Cycle Operations Unlock Enterprise Transformation

Transcript

How Agentic Revenue Cycle Operations Unlock Enterprise Transformation · Becker’s Healthcare Podcast

• 0:00
  - Hi, everyone. This is Chanel Bonger with Becker's
• 0:02
  - Healthcare. Thank you so much for tuning into
• 0:04
  - the Becker's Healthcare podcast series.
• 0:06
  - Today, we're gonna talk about how agentic revenue
• 0:08
  - cycle operations unlock enterprise transformation.
• 0:12
  - Joining me today is PV Subbarao, who leads
• 0:15
  - the healthcare business at NTT DATA in North
• 0:18
  - America with more than twenty five years of
• 0:20
  - experience across healthcare and life sciences. He focuses
• 0:24
  - on helping organizations
• 0:25
  - modernize, scale, and innovate while delivering strong revenue
• 0:28
  - growth and operational excellence.
• 0:30
  - Phoebe, thank you so much for joining me
• 0:32
  - today. Thank you so much. Good morning, and
• 0:34
  - glad to be here.
• 0:35
  - Likewise. Well, from an opening perspective, what is
• 0:38
  - your point of view on revenue cycle?
• 0:40
  - I think we are at an interesting point
• 0:42
  - in the evolution of the revenue cycle. For
• 0:45
  - many years, the focus has been on
• 0:47
  - digitization,
• 0:49
  - automation, and RPA,
• 0:50
  - and those have created
• 0:52
  - a significant value.
• 0:53
  - But, fundamentally, we are auto automating individual tasks
• 0:57
  - with an operating model that was still largely
• 0:59
  - dependent on people managing the work queues, navigating
• 1:02
  - multiple,
• 1:03
  - systems, and data mining what should happen next.
• 1:06
  - But what we see is that agentic AI
• 1:08
  - gives an opportunity to rethink that model.
• 1:10
  - We can move from task automation
• 1:13
  - toward outcome
• 1:14
  - oriented operations to where AI agents can understand
• 1:17
  - the context,
• 1:18
  - reason about what needs to happen next,
• 1:21
  - take actions across workflows and systems,
• 1:24
  - and involve people when judgment or intervention is
• 1:27
  - needed.
• 1:28
  - For health systems, the opportunity is bigger than
• 1:30
  - productivity, is about improving cash,
• 1:33
  - reducing administrative friction,
• 1:34
  - creating greater financial visibility,
• 1:37
  - and ultimately building a much more intelligent revenue
• 1:40
  - cycle.
• 1:41
  - Got it. And building on that, when you
• 1:42
  - talk about agentic operations and revenue cycle, what
• 1:45
  - does that look like day to day? And
• 1:48
  - what can an agentic model do that traditional
• 1:50
  - automation or RPA can't?
• 1:53
  - Very good question.
• 1:54
  - The simplest way I describe the difference is
• 1:57
  - this.
• 1:58
  - Traditional automation,
• 1:59
  - executes a task.
• 2:01
  - Agentical operations
• 2:03
  - pursue an outcome.
• 2:04
  - RPA has been very effective when,
• 2:06
  - the process is predictable.
• 2:08
  - Give it the rule, log in to system,
• 2:11
  - more information
• 2:12
  - from point a to point b, check a
• 2:15
  - claim status, and it can execute the task
• 2:17
  - thousands of times.
• 2:19
  - But the revenue cycle isn't always predictable. And
• 2:22
  - there are exceptions,
• 2:24
  - changing payrolls,
• 2:25
  - missing information,
• 2:27
  - multiple systems and situations where the,
• 2:30
  - next action depends on what happened previously.
• 2:33
  - This is where agent declarations become interesting.
• 2:36
  - Imagine a denied claim.
• 2:39
  - In in traditional model, automation
• 2:42
  - might identify
• 2:43
  - the denial,
• 2:44
  - categorize it, and put it in someone's work
• 2:46
  - queue. But in the agenting model, the objective
• 2:49
  - isn't simply to process the denial. The objective
• 2:52
  - is to get the claim resolved and get
• 2:54
  - paid.
• 2:55
  - And an agent can understand the denial reason,
• 2:58
  - look at the claim history, gather the relevant
• 3:00
  - information,
• 3:01
  - determine the appropriate next action,
• 3:04
  - prepare or initiate an appeal,
• 3:06
  - monitor the payer response, and and continue working
• 3:09
  - towards the solution. If it reaches a point
• 3:12
  - requiring clinical judgment or another human decision,
• 3:15
  - it can bring the right person into the
• 3:17
  - process with the relevant context already assembled.
• 3:20
  - That changes the role of the workforce.
• 3:23
  - Today,
• 3:24
  - highly capable people
• 3:26
  - spend an enormous amount of time
• 3:29
  - navigating through systems,
• 3:31
  - gathering information,
• 3:32
  - making phone calls,
• 3:34
  - checking status, and managing queues.
• 3:36
  - In an agentic model, machines increasingly handle that
• 3:40
  - administrative orchestration
• 3:42
  - while people focus on exceptions,
• 3:44
  - judgment,
• 3:45
  - payer strategy, higher value activities.
• 3:48
  - And then that's where they should, you know,
• 3:49
  - spend most of the time. And I think
• 3:51
  - that's an important distinction,
• 3:53
  - and that's about
• 3:55
  - removing humans from the revenue cycle.
• 3:57
  - It's about changing where human capacity is applied.
• 4:02
  - Over time, you can imagine the revenue cycle
• 4:04
  - becoming
• 4:05
  - less dependent on people,
• 4:07
  - constantly pushing transactions through the system,
• 4:10
  - and much more oriented around managing outcomes.
• 4:13
  - So simply put,
• 4:15
  - automation completes tasks, agentic operations pursue outcomes.
• 4:20
  - Got it. And digging a bit deeper on
• 4:22
  - that, where are you seeing agentic operations deliver
• 4:25
  - the fastest, most measurable results in the revenue
• 4:27
  - cycle today? And what makes those areas natural
• 4:30
  - starting points?
• 4:32
  - We see the strongest opportunities where three things
• 4:35
  - come together.
• 4:37
  - High transaction volume, significant administrative friction,
• 4:41
  - and clearly measurable
• 4:42
  - financial outcome.
• 4:44
  - So denials are an obvious example. So there
• 4:46
  - is a tremendous amount of work associated with
• 4:49
  - understanding
• 4:51
  - why a claim was denied,
• 4:53
  - gathering information, determining whether to appeal,
• 4:56
  - preparing that appeal,
• 4:58
  - submitting it, and and then following it through
• 5:00
  - the resolution.
• 5:02
  - A significant portion of that process is administrative
• 5:05
  - rather than, judgment based, which makes it very
• 5:09
  - well suited for
• 5:10
  - AI and agentic workflows.
• 5:13
  - But denials are only one area. Eligibility verification,
• 5:16
  - prior authorization,
• 5:18
  - claim status follow-up, underpayments,
• 5:21
  - payment variance,
• 5:22
  - AR follow-up, and and portions of patient financial
• 5:26
  - engagement are all areas where we see meaningful
• 5:29
  - opportunity.
• 5:30
  - What makes these,
• 5:31
  - particularly good starting points is that
• 5:34
  - economics are visible.
• 5:36
  - You can establish a baseline, before you start.
• 5:40
  - How many manual touches does it take to
• 5:42
  - resolve an account?
• 5:44
  - How long does it take to resolve?
• 5:46
  - What percentage of denials are recovered?
• 5:49
  - How much cash is,
• 5:50
  - sitting in here? What is our cash to
• 5:53
  - collect? So all of these metrics are important.
• 5:55
  - Then you can measure
• 5:56
  - whether the new operating model is actually changing
• 5:59
  - those outcomes.
• 6:00
  - And that's, important because,
• 6:03
  - there is a lot of excitement around AI
• 6:05
  - right now. Our systems don't need another technology
• 6:08
  - experiment.
• 6:09
  - They need measurable
• 6:11
  - business outcomes. So my advice is don't start
• 6:14
  - by asking, where can I deploy AI?
• 6:18
  - Start by asking,
• 6:20
  - where is value trapped in the revenue cycle?
• 6:23
  - Find few areas where there is a meaningful
• 6:26
  - frictional,
• 6:27
  - you know, considerations,
• 6:28
  - establish the baseline,
• 6:30
  - apply AI and agenty capabilities, and and prove
• 6:33
  - the economics.
• 6:34
  - So once you've demonstrated that, you have
• 6:37
  - both the business case and the organizational confidence
• 6:40
  - to scale. I also think,
• 6:42
  - there's an important progression,
• 6:45
  - here. Initially,
• 6:47
  - organizations may use agents within individual processes.
• 6:50
  - Over time, these, these particular agents
• 6:53
  - begin working across the revenue cycle continuum.
• 6:56
  - That's when the operating model itself starts to
• 6:58
  - change. So simply put, start with the where
• 7:02
  - the friction is high, volume is high, and
• 7:04
  - financial value is measurable.
• 7:07
  - Got it. I love how you frame that.
• 7:08
  - And you kinda touched on cash there a
• 7:10
  - bit. And digging it to a bit deeper
• 7:12
  - on that, How do improvements in cash position
• 7:15
  - and margin from the revenue cycle translate into
• 7:18
  - greater freedom to invest in transformation
• 7:21
  - elsewhere across the enterprise?
• 7:24
  - This is where I think the revenue cycle
• 7:26
  - conversation becomes much more strategic. Health systems are
• 7:29
  - under enormous pressure, and we've been seeing that.
• 7:32
  - And they are trying to improve margins,
• 7:35
  - while simultaneously investing in clinical transformation,
• 7:39
  - workforce transformation, cybersecurity,
• 7:41
  - digital
• 7:43
  - capabilities, and data and AI.
• 7:44
  - These investments
• 7:46
  - aren't optional,
• 7:47
  - but the capital to fund these initiatives
• 7:50
  - has to come from somewhere. So the revenue
• 7:53
  - cycle is one of the places where significant
• 7:55
  - value can be unlocked.
• 7:57
  - Every dollar we collect
• 7:59
  - that otherwise might have been written off,
• 8:01
  - every day, we reduce in accounts receivable.
• 8:05
  - Every unnecessary administrative touch we eliminate, and every
• 8:08
  - improvement in cost to collect, you know, contributes
• 8:11
  - to the financial capacity of the organization.
• 8:14
  - So I don't think of revenue cycle transformation
• 8:16
  - simply as a back office efficiency program.
• 8:20
  - Done well, it can become a self funding
• 8:22
  - transformation, you know, for the entire enterprise.
• 8:25
  - And, you improve cash, you know, velocity,
• 8:28
  - and you reduce leakage. You lower structural operating
• 8:31
  - costs. And some of that value can then
• 8:34
  - be reinvested into broader transformation agenda.
• 8:36
  - So that could mean funding a clinical modernization
• 8:40
  - initiative,
• 8:41
  - accelerating cloud or data transformation,
• 8:44
  - strengthening cybersecurity,
• 8:45
  - investing in AI, or improving
• 8:48
  - the patient and, clinician experience.
• 8:51
  - And and,
• 8:52
  - there is another benefit that sometimes gets overlooked,
• 8:55
  - management capacity.
• 8:56
  - If your leadership teams and workforce are constantly
• 8:59
  - dealing with
• 9:01
  - operational friction today,
• 9:03
  - handling exceptions and escalations,
• 9:05
  - they may have less capacity to think about
• 9:08
  - the true transformation.
• 9:09
  - So when you create more autonomous revenue cycle
• 9:12
  - operation, you're potentially freeing up both financial capacity
• 9:16
  - and also human capacity.
• 9:18
  - So that's why I believe CFOs should view
• 9:20
  - this as
• 9:21
  - a more than RCM initiative.
• 9:24
  - It can become part of the broader enterprise
• 9:26
  - transmission strategy.
• 9:28
  - Simply put,
• 9:29
  - a better revenue cycle doesn't just improve the
• 9:32
  - back office. It create investment capacity for the
• 9:35
  - enterprise.
• 9:37
  - Absolutely. That makes a lot of sense. And
• 9:39
  - now one of the most underappreciated
• 9:41
  - outputs of agentic operations is greater visibility.
• 9:44
  - What can CFOs and revenue cycle leaders see
• 9:47
  - about financial performance today that they couldn't see
• 9:50
  - two years ago? And how is that changing
• 9:52
  - the decisions that they
• 9:54
  - make? Very loaded question.
• 9:56
  - I think this may ultimately be as important
• 9:59
  - as, the automation itself.
• 10:02
  - Historically,
• 10:03
  - health care, has had a tremendous amount of
• 10:05
  - revenue cycle data. Now the challenge
• 10:08
  - hasn't necessarily
• 10:10
  - been the absence of data. It has been
• 10:12
  - turning that data into timely,
• 10:15
  - actionable intelligence.
• 10:17
  - Most traditional dashboards are very good at telling
• 10:21
  - us what happened.
• 10:22
  - What was our cash collection?
• 10:24
  - What are our days in AR? What is
• 10:27
  - our denial in our rate? What's our cost
• 10:30
  - to collect? These are all important measures, but
• 10:32
  - they are largely retrospective.
• 10:35
  - But what AI brings to give us is
• 10:37
  - the ability to understand
• 10:39
  - why something is happening,
• 10:41
  - what is likely to happen next, and what
• 10:44
  - action we should take. For example, imagine being
• 10:47
  - able to see that a particular pair
• 10:49
  - denial behavior is beginning to change,
• 10:52
  - before it becomes a major financial issue
• 10:55
  - or recognize that a certain combination of procedure,
• 10:59
  - documentation, and payer
• 11:01
  - is creating a higher probability of a denial
• 11:04
  - and intervening
• 11:05
  - before the claim itself is submitted
• 11:08
  - or identifying which accounts
• 11:10
  - in a very large AR portfolio
• 11:13
  - have the highest probability of recovery and automatically
• 11:16
  - prioritizing the work accordingly.
• 11:18
  - That's where it's a very different management model
• 11:21
  - that you can think of.
• 11:24
  - Instead of saying our denial rate went up
• 11:26
  - last month, let's investigate,
• 11:28
  - usual conversation. But you begin moving forward,
• 11:32
  - with a with an idea of, we are
• 11:34
  - seeing a pattern emerging
• 11:36
  - with this particular payer. Here is the likely
• 11:39
  - financial exposure,
• 11:41
  - and here are underlying drivers, and here are
• 11:43
  - the actions we should take now.
• 11:46
  - So for the CFO, that creates a much
• 11:48
  - stronger connection between operational activity
• 11:52
  - and the financial forecasting.
• 11:54
  - And as, agentic systems, execute more of the
• 11:57
  - work, you also get visibility into the workflow
• 12:00
  - itself in the system where actions were taken,
• 12:04
  - where processes are
• 12:05
  - getting stuck, which interventions are working, and where
• 12:09
  - human judgment is consistently required.
• 12:12
  - That creates a continuous feedback loop. So we
• 12:15
  - are moving from reporting
• 12:17
  - to intelligence and ultimately from intelligence to action.
• 12:21
  - I think that will fundamentally change how revenue
• 12:23
  - cycle leaders, you know, manage performance.
• 12:26
  - In in summary, the future isn't another dashboard.
• 12:29
  - We've got many of them, in today's world.
• 12:32
  - It's knowing
• 12:33
  - what happened,
• 12:35
  - why it happened,
• 12:37
  - what is likely to happen next, and what
• 12:39
  - to do about it.
• 12:41
  - Got it. Got it. Well, Pee Wee, you've
• 12:42
  - already shared a lot of actionable,
• 12:44
  - advice. But while I've got you, I'm gonna
• 12:46
  - ask for a little more. For health systems
• 12:48
  - looking to move towards agentic operations,
• 12:51
  - what foundational capabilities need to be in place
• 12:54
  - first, and where should they start?
• 12:56
  - The first thing I would say is don't
• 12:58
  - make this primarily a technology project. You know,
• 13:01
  - technology obviously matters, but successful agentic operations require
• 13:05
  - several foundations.
• 13:07
  - The first is data
• 13:08
  - and connectivity.
• 13:10
  - Agents need reliable access to the information and
• 13:13
  - systems
• 13:14
  - required to do the actual work. That doesn't
• 13:17
  - mean every health system needs to completely modernize
• 13:19
  - its technology estate before starting any type of
• 13:22
  - agentic RCM work. But the agents needs access
• 13:25
  - to the right context
• 13:27
  - in in their enterprise.
• 13:28
  - So second is workflow understanding. So before applying
• 13:31
  - AI,
• 13:32
  - you need to understand how work actually gets
• 13:35
  - done today.
• 13:36
  - Not simply how process was designed on paper.
• 13:39
  - Where are the handoffs? Where are the exceptions?
• 13:42
  - Where do people spend their time?
• 13:45
  - Where does value leak out of the process?
• 13:48
  - And then the third is governance and trust.
• 13:51
  - In in health care, we have to be
• 13:53
  - very thoughtful about what an agent is authorized
• 13:56
  - to do automatically,
• 13:58
  - what needs validation,
• 14:00
  - where a human needs to remain in the
• 14:02
  - loop.
• 14:03
  - That means
• 14:04
  - auditability,
• 14:06
  - security,
• 14:07
  - appropriate controls,
• 14:09
  - and and clearly defined accountability are very much
• 14:12
  - critical.
• 14:13
  - And fourth is economics.
• 14:15
  - You need to know the baseline
• 14:17
  - before transforming,
• 14:18
  - any particular process.
• 14:20
  - Otherwise,
• 14:21
  - six months later, everyone may agree that the
• 14:24
  - technology is super impressive,
• 14:26
  - but nobody can clearly demonstrate
• 14:28
  - whether the initiative itself has created value or
• 14:31
  - not.
• 14:32
  - So where should organizations
• 14:34
  - start? So if you look at it from
• 14:36
  - that lens, I would choose two or three
• 14:38
  - high value use cases, not 20.
• 14:40
  - Pick areas with meaningful
• 14:43
  - volume
• 14:44
  - and friction
• 14:45
  - where the outcome is measurable,
• 14:48
  - establish a baseline,
• 14:49
  - introduce agentic capabilities
• 14:51
  - with appropriate human oversight,
• 14:54
  - measure the impact, learn from it, and then
• 14:56
  - scale.
• 14:57
  - And as you, scale, don't simply layer agents
• 15:01
  - on top of every existing workflow.
• 15:03
  - At some point, you have to ask, if
• 15:05
  - I were designing this revenue cycle operation today
• 15:08
  - with AI agents
• 15:10
  - available from the beginning, would I design it
• 15:12
  - the same way? And and in many cases,
• 15:14
  - the answer will be no. And that's where
• 15:16
  - the real transformation occurs. The the the designation
• 15:19
  - isn't a revenue cycle with hundreds of bots
• 15:22
  - doing the same work people used to do,
• 15:24
  - historically.
• 15:26
  - The the destination is fundamentally a different operating
• 15:28
  - model where people,
• 15:30
  - AI agents, automation, and data work together
• 15:33
  - with humans
• 15:35
  - increasingly focused on judgment,
• 15:37
  - relationships,
• 15:38
  - and exceptions,
• 15:39
  - and machines handling more of the transactional complexity.
• 15:42
  - Simply put, don't automate the old operating model.
• 15:46
  - Use agentic AI as an opportunity to redesign
• 15:49
  - it.
• 15:50
  - Got it. Got it. Well, PV, I wanna
• 15:52
  - thank you for your time today. But before
• 15:53
  - I let you go, is there anything else
• 15:55
  - that listeners should know?
• 15:57
  - As a closing comment or anything that I
• 15:59
  - would like to share, it is more from
• 16:01
  - a historic perspective of what, I have seen
• 16:04
  - health systems
• 16:05
  - been grappling with. You know, as if I
• 16:07
  - step back, the promise of agentic operations isn't
• 16:11
  - simply about doing today's revenue cycle work faster
• 16:14
  - or cheaper.
• 16:15
  - It's about creating
• 16:17
  - a revenue cycle that is more autonomous,
• 16:20
  - more intelligent, and increasingly predictive.
• 16:23
  - One that doesn't simply process
• 16:26
  - bunch of transactions but continuously work towards outcomes.
• 16:30
  - Getting the right claim paid, identifying financial leakage
• 16:33
  - earlier,
• 16:35
  - prioritizing the right intervention, and giving
• 16:38
  - leaders much greater visibility into what is happening
• 16:41
  - across the business. And if we do that
• 16:43
  - well, there are really three difference.
• 16:46
  - We improve cash and margin.
• 16:48
  - We reduce administrative burden and free up human
• 16:50
  - capacity.
• 16:52
  - And the third one is we create greater
• 16:54
  - financial capacity for the health system to invest
• 16:56
  - in transformation.
• 16:58
  - And that's the connection I think is most
• 17:00
  - important.
• 17:01
  - So revenue cycle transformation isn't an end in
• 17:04
  - itself.
• 17:05
  - If we can remove administrative fraction
• 17:07
  - and unlock financial capacity, we give health systems
• 17:11
  - greater freedom to invest in their clinicians,
• 17:14
  - their patients, innovation, and the communities that they
• 17:17
  - serve. That to me is the, real promise
• 17:20
  - of an AI NATO agentic revenue cycle.
• 17:23
  - Phoebe, that's a great takeaway to end on.
• 17:25
  - I wanna thank you once again for your
• 17:27
  - time today. And we also wanna thank our
• 17:29
  - podcast sponsor, NTT DATA. You can tune into
• 17:32
  - more podcasts from Becker's Healthcare by visiting our
• 17:35
  - podcast page at beckershospitalreview.com.
• 17:37
  - Thank you for having me here today.