Latest / Becker’s Payer Issues Podcast / Harnessing AI to Deliver Breakthrough Healthcare Value, Faster

Transcript

Harnessing AI to Deliver Breakthrough Healthcare Value, Faster · Becker’s Payer Issues Podcast

• 0:00
  Hi, everyone. This is Brian Zimmerman with Beckerts
• 0:02
  Healthcare. Thank you so much for tuning into
• 0:04
  the Beckerts Healthcare podcast.
• 0:06
  Today, we're going to talk about harnessing AI
• 0:09
  to deliver breakthrough health care value faster.
• 0:12
  Joining me for today's discussion is Sundar Srinivasan,
• 0:15
  president health plan and life sciences with NTT
• 0:19
  DATA. Sundar, thank you so much for being
• 0:21
  here today. Thank you, Brian. Happy to be
• 0:22
  here. Wonderful. So to to get started here,
• 0:25
  could can you just share a bit more
• 0:27
  about your your professional background and the work
• 0:29
  you're doing now?
• 0:30
  Absolutely.
• 0:32
  I've been serving the health care clients for
• 0:34
  over twenty five years.
• 0:36
  But in my current role, I lead the
• 0:38
  health plan and life sciences business at NTT
• 0:41
  DATA,
• 0:42
  where I work closely with a little over
• 0:45
  50 health plans, including
• 0:47
  large national payers, blues, and regional plans of
• 0:50
  varying scale.
• 0:52
  See the vantage plan this gives me
• 0:55
  is that it gives me a broad perspective
• 0:57
  on how the industry is evolving,
• 1:00
  especially as organizations
• 1:02
  navigate cost pressures, regulatory complexity,
• 1:05
  and rising expectations around member and provider experience.
• 1:10
  A big part part of my focus is
• 1:12
  helping these plants
• 1:14
  translate
• 1:14
  emerging technologies like AI, particularly
• 1:17
  agentic AI,
• 1:18
  from early experimentation
• 1:20
  into real
• 1:22
  scaled impact across areas like claims, prior authorization,
• 1:26
  and customer engagement.
• 1:28
  Excellent, Hidar. I will let's let's pull on
• 1:30
  that expertise and that experience. So I believe
• 1:33
  that a a recent Deloitte report found that
• 1:36
  eighty five percent of health care leaders plan
• 1:38
  to increase investment in the at AgenTic AI
• 1:40
  over the next few years. I think if
• 1:42
  you talk to a lot of folks,
• 1:44
  in in the industry,
• 1:45
  it suggests that that share of folks is
• 1:47
  even gonna increase from from the 85% number
• 1:50
  over time.
• 1:51
  From where you sit, though, can you just
• 1:52
  give us a a a lay of the
• 1:53
  land here in terms of how this trend
• 1:55
  is playing out in the payer space specifically?
• 1:58
  What is really driving the the momentum behind
• 2:01
  this investment?
• 2:02
  Absolutely. It's absolutely true. We are definitely
• 2:07
  seeing that trend
• 2:08
  accelerate in the payer space.
• 2:11
  And in many cases, payers are moving faster
• 2:13
  than providers, at least from my vantage point.
• 2:16
  The big driver is that agentic AI
• 2:20
  directly targets
• 2:21
  administrative
• 2:22
  heavy workflows like claims, prior auth,
• 2:25
  and contact centers to name a few. There
• 2:27
  are others as well, but those are the
• 2:29
  primary ones they're focusing.
• 2:31
  What is different now is the shift from
• 2:34
  simple automation to end to end orchestration.
• 2:39
  It AI agents can actually execute multi step
• 2:41
  processes today. I mean, the technology is amazing
• 2:44
  what what's out there,
• 2:46
  which creates a much clearer return on investment
• 2:49
  for payers.
• 2:50
  For payers, that means lower cost per claim,
• 2:54
  faster turnaround times,
• 2:56
  and fewer manual touches. In essence,
• 2:59
  you're getting better provider and member experience in
• 3:02
  the
• 3:03
  end. What's,
• 3:05
  at the same time, there are barriers like
• 3:07
  data readiness and leadership alignment
• 3:10
  that are
• 3:11
  improving. So organizations
• 3:13
  are moving beyond pilots
• 3:15
  and into deployments.
• 3:17
  And there is real competitive pressure.
• 3:19
  Early adopters are pulling ahead,
• 3:22
  which is pushing
• 3:23
  others to invest quickly.
• 3:26
  So this isn't experimental
• 3:28
  anymore.
• 3:29
  Agentic AI is becoming core to how payers
• 3:32
  rethink their operating model.
• 3:34
  Lot of them are transforming
• 3:36
  to a more platform based business process as
• 3:39
  a service, the smaller ones.
• 3:41
  Or even sometimes the newer term that we
• 3:43
  use is
• 3:44
  service as a software model.
• 3:47
  We use we are used to the software
• 3:49
  as a service, but that's flipping to service
• 3:52
  as a software model.
• 3:54
  Investment in is accelerating because agent
• 3:58
  AI hits the payer sweet spot.
• 4:00
  It reduces administrative
• 4:02
  cost, improves
• 4:04
  speed index experience,
• 4:06
  and creates a path to operating model transformation.
• 4:10
  And now that the barriers are lower
• 4:13
  and ROA is clearer,
• 4:15
  organizations
• 4:16
  are scaling fast
• 4:18
  rather than experimenting.
• 4:20
  That's what I see today. Right. Yeah. And
• 4:22
  and let's get into how this technology has
• 4:25
  matured sort of your your point around moving
• 4:27
  from
• 4:28
  simple automation to to end to end execution
• 4:31
  or this this platform strategy.
• 4:34
  As this maturity
• 4:35
  sort of goes on,
• 4:37
  what are the use cases that are rising
• 4:39
  to the top for payer organizations? I think
• 4:41
  you've gave a good overview of where,
• 4:43
  this technology can make a difference in terms
• 4:45
  of results you can drive, but let's let's
• 4:47
  zero in and give get even more specific.
• 4:49
  Which ones are you seeing generate the most
• 4:50
  excitement,
• 4:51
  and the most tangible value right now?
• 4:54
  Good question. That's what,
• 4:56
  everybody has when we talk to Peir, so
• 4:58
  it's good to talk about that.
• 5:00
  We are seeing a few use cases
• 5:03
  consistently
• 5:03
  rise to the top for payers.
• 5:06
  Primarily,
• 5:07
  areas where there is high volume,
• 5:10
  complexity,
• 5:11
  and administrative
• 5:12
  burden.
• 5:13
  I'll there are subtle differences in the kinds
• 5:15
  of, use cases, and I'll highlight some of
• 5:18
  them.
• 5:19
  Claims management is a big one.
• 5:21
  Intake
• 5:22
  adjudication support and exception handling, where agentic AI
• 5:26
  reduces manual touch points,
• 5:29
  speed up cycle times, and improves accuracy.
• 5:32
  That's number one. Every because there is not
• 5:35
  a lot of regulatory things there. It's a
• 5:38
  little easier with administrative,
• 5:40
  so people tend to do that, number one.
• 5:43
  Number two use case that I'm seeing, which
• 5:45
  they are trying to do this, but there
• 5:47
  is a little bit of regulatory
• 5:49
  pressure here. So they are cautious about doing
• 5:51
  this, and that area is prior authorization.
• 5:55
  So prior authorization
• 5:56
  and appeals processes.
• 5:58
  The second I would I'd say that would
• 6:00
  come to my mind right now,
• 6:02
  especially around documentation gathering, decision support,
• 6:06
  and enhanced explainability.
• 6:09
  And then the last third piece, I would
• 6:11
  say, is the contact center operations.
• 6:14
  AI agents assisting
• 6:15
  or resolving
• 6:16
  member and provider
• 6:18
  inquiries end to end.
• 6:20
  What makes these exciting is
• 6:22
  they're not just efficiency
• 6:24
  place. They directly impact cost,
• 6:27
  provide an expiration, and as I said earlier,
• 6:30
  member experience as well. So it's not just
• 6:32
  cost. It's the experience that they are looking
• 6:36
  for for their stakeholders.
• 6:38
  Those are some true truly exciting use cases.
• 6:42
  But but I do imagine, Sundar, that there
• 6:44
  are multiple vendors that could
• 6:46
  or at least say that they can make
• 6:48
  these use cases a reality.
• 6:50
  I I I guess my question for you
• 6:51
  in this this crowded technology space, how do
• 6:54
  you guide health plan leaders through evaluating and
• 6:56
  adopting these tools?
• 6:57
  What are some best practices or strategies that
• 7:00
  you share with your partners to to help
• 7:02
  them confidently get the most out of AgenTic
• 7:04
  AI?
• 7:05
  Yeah. Before I answer the question, there are
• 7:07
  other use cases as well.
• 7:09
  So those three were just the top of
• 7:11
  the mind use cases, so we'll certainly there
• 7:13
  are other use cases.
• 7:14
  But, to answer your specific question,
• 7:18
  on the adoption side,
• 7:20
  we guide plans to start with very targeted
• 7:24
  high return on investment workflows
• 7:26
  rather than developing a vision for broad transformation,
• 7:30
  because it gets very complex and, you know,
• 7:32
  payers tend or any enterprise tends to lose
• 7:35
  out on that.
• 7:37
  So our guide guidance is pick a small
• 7:39
  set of use cases
• 7:41
  that you can measure impact quickly.
• 7:44
  Cost per transaction,
• 7:45
  turnaround time, or call deflection.
• 7:48
  One thing some payers
• 7:51
  adopt is fail fast. They'll try something very
• 7:54
  quickly.
• 7:55
  They'll fail fast. If that's not the one
• 7:57
  they want to do, then they move to
• 7:58
  the next one. But then pick the areas
• 8:01
  that they really want to do and really
• 8:04
  stick to,
• 8:05
  proving that area. When I say fail fast,
• 8:07
  it's about the technology they use rather than
• 8:10
  the use case.
• 8:11
  Then design with a human in the loop
• 8:14
  model upfront.
• 8:16
  So there is trust and governance
• 8:18
  and scale from there.
• 8:20
  The other big piece is integration. Right? Embedding
• 8:24
  agents into existing workflows rather than layering them
• 8:27
  on top. This is very, very key
• 8:29
  in how you design this correctly.
• 8:32
  The plans that are getting the most value
• 8:35
  are treating this as an operating model ship
• 8:38
  and not just a technology deploy deployment,
• 8:42
  and building the roadmap
• 8:44
  accordingly.
• 8:45
  Lot of payers realize that it is important
• 8:48
  to understand that agentic AI
• 8:50
  offers the ability to reimagine business processes,
• 8:54
  workflows,
• 8:55
  and the role of team members.
• 8:58
  And when I talk about team members, a
• 9:00
  lot of times, you know, AI people are
• 9:02
  concerned about
• 9:03
  AI coming in, but the role of of
• 9:06
  the employee shift within an organization. So you
• 9:08
  need that
• 9:10
  transparency
• 9:11
  with the teams.
• 9:12
  To build that vision requires early successful momentum,
• 9:16
  an organization that's ready to embrace the change,
• 9:19
  and you need a bold vision from the
• 9:21
  payers.
• 9:23
  Yeah. This operating model ship, this this bold
• 9:25
  vision, I think you've really made the case
• 9:28
  about why this is the the correct strategy,
• 9:30
  but putting it in play can can also
• 9:33
  be a challenge. You need to make some
• 9:35
  organizational changes to make this happen. So I
• 9:37
  guess
• 9:38
  how do health plan leaders really build, I
• 9:41
  could the perhaps the organizational muscle or even
• 9:44
  in the case of larger organizations,
• 9:46
  even the the proper agility
• 9:48
  to scale and and make a credible case
• 9:50
  for AI's value across the enterprise? What's, I
• 9:53
  guess, one step a leader could take now
• 9:55
  to make that organizational change, that that readiness
• 9:58
  for AI a reality?
• 10:01
  Yeah. There are two types of,
• 10:03
  scale. One is, you know, you take one
• 10:06
  transaction and you scale it, to a large
• 10:08
  extent.
• 10:09
  The second is really, how do you pro
• 10:11
  proliferate
• 10:12
  the number of use cases so you can
• 10:14
  have more use cases running agenting?
• 10:17
  So to address both, the biggest shift from
• 10:20
  pilots to scale
• 10:21
  is treating AI as a business capability,
• 10:25
  not a series of experiments.
• 10:27
  The plans that succeed are building muscle around
• 10:30
  three things,
• 10:32
  clear ownership,
• 10:33
  repeatable governance,
• 10:35
  and value measurement.
• 10:37
  That means
• 10:38
  aligning AI initiatives
• 10:40
  to become business outcomes, like cost per claim
• 10:44
  or prior auth turnaround,
• 10:46
  so you can consistently
• 10:47
  prove your return on investment and then reinvest
• 10:51
  as you see success.
• 10:53
  It also means
• 10:54
  creating a cross functional model
• 10:57
  where business,
• 10:58
  operations and IT are jointly
• 11:01
  accountable
• 11:01
  and not working in silos.
• 11:03
  The challenge we see a lot of times
• 11:05
  is teams work in silos, the IT teams
• 11:08
  work one way, the operating teams, operations team,
• 11:12
  teams work in another way,
• 11:14
  there is not a lot of success. So
• 11:15
  if they're working jointly
• 11:17
  and the accountability lies together,
• 11:19
  then we see a lot of success. If
• 11:21
  there is one step leaders can take now,
• 11:24
  it's this.
• 11:26
  If one high impact workflow and fully operationalize
• 11:30
  it end to end with metrics,
• 11:32
  governance,
• 11:33
  and change management built in. Because once you
• 11:37
  prove value in one area,
• 11:39
  scaling becomes a lot easier and more importantly,
• 11:42
  a lot more credible, Brian.
• 11:45
  Yeah. And and to your point, it's gotta
• 11:47
  be high impact. Right? It's it's gotta be
• 11:48
  a place that's gonna drive meaningful change. Correct,
• 11:51
  Sundar? Am I Absolutely, Brian. Absolutely.
• 11:54
  Excellent.
• 11:55
  Before I let you go, I wanna give
• 11:57
  you opportunity either,
• 11:59
  share a final parting takeaway or perhaps,
• 12:02
  you you know, share something or or or
• 12:04
  add some commentary around
• 12:06
  a a portion or an element of this
• 12:08
  conversation that you think deserves more attention,
• 12:12
  than than it got in our conversation today.
• 12:15
  Yeah. See, one thing that deserves
• 12:18
  a lot more attention in my view is
• 12:20
  trust and governance. We are in, at all,
• 12:22
  in a regulated industry.
• 12:24
  And
• 12:25
  payers are typically discoverers when it comes to
• 12:29
  adopting these new technologies.
• 12:31
  But,
• 12:32
  today, the technology is so powerful
• 12:35
  that they are forced to take decisions plus
• 12:38
  the industry is going through a lot of,
• 12:41
  you know, challenges today.
• 12:43
  So an agent decay is giving them an
• 12:45
  opportunity
• 12:46
  to really address some of their challenges.
• 12:49
  So as agent decay takes
• 12:51
  on a more autonomous decision making
• 12:55
  capability, especially
• 12:56
  in areas like claims or prior auth.
• 12:59
  Payers have to be very intentional about transparency,
• 13:03
  explainability,
• 13:05
  auditability,
• 13:06
  and guided decision making.
• 13:09
  The technology is moving very fast,
• 13:11
  but scaling it responsibly
• 13:14
  is what ultimately determine
• 13:16
  long term, success. We use a phrase called
• 13:19
  responsible
• 13:20
  AI. It's important to,
• 13:23
  keep that in mind.
• 13:24
  The organizations
• 13:26
  that get this right
• 13:28
  won't just see efficient
• 13:30
  efficiency gains, Brian.
• 13:31
  They'll build trust with providers, members, and regulators,
• 13:35
  which is just as critical. At the end
• 13:37
  of the day,
• 13:38
  the payers are serving
• 13:40
  the members. And together between the providers and
• 13:43
  payers, they need to make sure that the
• 13:45
  well-being of the,
• 13:47
  members are taken care of. And
• 13:50
  agentic AI and the technologies are ultimately going
• 13:53
  to help. So having more transparency
• 13:55
  helps build that trust and then the better
• 13:58
  experience they can get out of it. Right?
• 14:01
  Yeah. Fascinating to think about this technology as
• 14:03
  a potential,
• 14:05
  trust builder across stakeholders. Right, Sundar?
• 14:08
  Absolutely. Yes. Trust is the most thing because
• 14:11
  whenever AI is involved, people are they're kind
• 14:15
  of scared about AI. Is it going to
• 14:17
  make the right decision? But the interesting thing
• 14:19
  is that based on statistics,
• 14:22
  actually humans make more errors than AI. So
• 14:25
  to prove that the AI has not made
• 14:28
  the mistake
• 14:29
  is very critical. So once you build the
• 14:31
  trust,
• 14:32
  clearly, you know, you have a lot more
• 14:33
  opportunities to scale
• 14:35
  both the number of use cases as well
• 14:37
  as the volume that you can throw out
• 14:39
  to on those use cases.
• 14:41
  Excellent. Well, well, Sundar, it was a pleasure
• 14:44
  speaking you with you today. Thank you so
• 14:45
  much for coming on the podcast.
• 14:47
  Thank you, Brian. Thank you for having me.
• 14:49
  We also wanna thank our podcast sponsor, NTT
• 14:52
  DATA. You can tune in to more podcasts
• 14:54
  from Becker's Healthcare by visiting our podcast page
• 14:57
  at beckershospitalreview.com.