Subscribe
Share
Share
Embed
Where we share our weekly news debriefs and discussions with industry experts. These are lo-fi recordings aimed at giving our readers more opportunities to engage with our analysis and a view into some of the conversations that shape it.
Martin: hey Neil.
Neil: Hey guys.
How are you doing?
Martin: Good.
How are you today?
Neil: Oh, I'm good.
Sorry.
I'm good.
Uh, just, I'm, I'm middle of moving and
so when I'm middle of moving, so I wanna
make sure that I got, uh, Internet's good.
Um, it's like a chaotic time out here.
I'm trying to hide all the boxes.
Yeah,
Martin: no, you look great.
Um, I think a good place to start
would be just like a quick update
on the pair team approach today.
For viewers and listeners
who aren't as familiar,
Neil: I.
Yeah, of course.
One.
Thank you guys for having, having us on.
Um, so for folks that Dunno, pair
Team is an AI powered care management
group and we focus on individuals
who, whose health first starts with
housing, food, transportation, and
other kind of basic life needs.
And so we provide both system
navigation across our healthcare
and public assistance programs.
Um, but we are also a
full stack medical group.
So we are a full stack virtual medical
group with nps, RNs, behavioral
health specialists, so we can kind
of manage, um, the end, you know,
manage a patient, patient from any
sort of health or life need they have.
Um, and it's been, uh, a ton of fun.
I think things like access
make this pretty exciting.
Um, and I know we were gonna
talk about that a little bit.
Kevin: Yeah.
Neil, give us a so, so access, you
also wrote paper recently on AI and
Medi-Cal and the opportunity there.
Give us a sense of just.
Scope and scale of, of
pair team business today?
When you say medical group,
is it virtual medical group?
Are you in person?
What markets are you in?
How do you, how do you
think about the world?
Neil: Yeah.
Yeah.
So we started, started this
organization and focused on a very
high needs patient population.
So focused on individuals experiencing
homelessness, severe mental illness,
substance use, um, poly chronic folks
that are frequenting the emergency room.
And we focused in high needs Medicaid,
particularly in a program, um, that came
out in California called Enhanced Care
Management, which was just super aligned.
And one of the problems here that you
see is, um, the handoff problem where
someone, you know, a, uh, this fellow Cody
Kinsley, uh, North Carolina State Health
Department, uh, state Medicaid director
used this term pamphlet and pray, which
is someone walks into the emergency room
and you go, well, hey, here's the pamphlet
to get to the resources that you need.
Um, do they get there?
No one knows, right?
And then you have platforms like Find
Help and United, which help like, create
all the connectivity, but even then
it's still hard to like actually close
the loop from a services perspective.
And so where Pair team has been,
is focused on that high needs
population in a very full stack.
Approach.
So we have this virtual medical
group, but we also build out a direct
network of contracted, um, social
care providers and medical providers.
So we'll partner with libraries, with
food pantries, with sobering centers,
with, um, YMCAs, all sorts of different
community-based organizations that
are integrated into our model so that
you can provide and get, you know,
you reduce that handoff risk, you can
actually get someone to their care.
Um, uh, last year we put out a paper.
Uh, this was a, you know, showing
our outcomes, 52% ed reduction,
26% inpatient reduction.
It showed the model really working.
If you have community health workers,
virtual medical group in this last
Mile Care Delivery network, um, about
nine months ago, and Kevin, I'll get
to the, the paper on ai, but just to
give folks the, the story about nine
months ago, you know, our mission
has always been Whole Person care for
everyone, not just whole Person Care,
you know, for a high needs population.
And to do that, you really
have to lower cost to serve.
I think everyone knows what
Medicaid rates look like.
It is for someone to go through the
Medicaid system is, is truly heroic.
It is incredibly fragmented.
You have all these life needs, you know,
kind of working up, trying to work up
Maslow's hierarchy of needs for yourself.
Um, and you gotta do it alone.
How do we bring our personal care
management touch to everyone been
focusing on AI and what that can do?
And about nine months ago we had
this, what I'd call like a, a
watershed moment for us where.
We had built a first version of
Flora, who is our AI care advocate,
kind of trained her with all
of our protocols that we'd had.
Um, and she had a voice call
with, um, an individual.
This was a senior living out of her
car, PTSD, congestive heart failure.
And it was over an hour
long phone conversation.
It was way deeper, way more rich
than, um, we could have expected.
And like both you go,
wow, the tech is here.
It's also kinda heartbreaking.
At the same time in that moment of
just realizing kinda like the social
isolation that that happened, uh, you
know, that's so, it's so pervasive.
Um, and so as soon as we saw that
moment, we said, okay, we're, we're here.
This is our way to get whole
person care to everyone.
And so what we've been building is
Flora, who is like I I mentioned our
AI advocate who can help engage with
someone over voice and text and not
just be that 24 7 care manager, but
also be a companion to them, but still
backed with our full care model, full
backed by our virtual medical group,
our last mile Care delivery network.
It's not just chat GPT about health.
Right.
Um, um, and so as part of that, we entered
into, you know, we've been talking to
CMS and CMI about the ACCESS program
and we are, we are now officially an
access participant on bringing that Whole
Person Care model to the safety net.
We believe that care management requires
not just health navigation, but also
public assistance navigation as well.
We've been building out this model forever
and Flora can now drive it and be the
core kind of interface for it to bring
it to a much broader patient population.
Um, and as part of this, we've been
really investing in, uh, obviously
safety effectiveness of Flora.
It is now a typical, it's not
typical for us to have 45 minutes,
60 minute long conversations with
seniors who Flora is actively caring
for with support of our care team.
Um, and um, alongside that we've
also just been very deep on
infrastructure for the safety net.
I think when you see things like, you
know, when, um, uh, EMRs came out,
you saw this technology typically
kind of gloss over the safety net.
You still have FQHCs that are
using very outdated tools relative
to commercial counterparts.
And we wanna make sure
that doesn't happen here.
So we're thinking about things, um, like,
uh, if you are a state health department,
if you are in the federal seat,
building out Medicaid IT infrastructure,
how do you make it AI native.
How do you, uh, I can go a little bit
into, into that, but you know, we put out
papers on this is what our spec would look
like and do everything as open source as
possible, but incorporates things like,
let's break down the, the data silos
with, you know, model context, protocol
interfaces, you know, for that are kind of
more agent ready and um, you know, consent
systems so that people have control
over their data and kind of own it.
Um, that's a very important part to all
this, but that's, that's been pair team
journey in AI is we've set, we started
this whole person care model for high
needs individuals to bring it to ev had
our watershed moment and now bringing
it to, um, everyone in the safety net
with, with Flora, our AI care advocate.
Martin: I'd be curious
to hear how you process.
It feels like there's kind
of two camps emerging.
One on the side of AI saying,
Hey, it is, uh, this is too big
of a challenge to not use AI.
And, you know, so we had, um, Andy from
Ol Ventures and Toyen from C Block talking
about, you know, the need to, to sort
of really engage with AI to support.
Folks, um, who are enrolled in Medicaid.
And then the other side, you, you were
starting to see some pushback from folks
like the, the Medical Board of Utah,
um, on the doct chronic pilot basically
saying, this is moving too quickly.
The incentives are bad.
We, we shouldn't be treating
patients like Guinea pigs, some,
you know, some other corners.
Here's how you process that debate and how
you are thinking about those questions.
Neil: Yeah, yeah.
Um, also I realized, Kevin, I didn't
answer we're, we are coming up on a
thousand person medical group right now.
Wow.
So we've gotten to a pretty decent,
I would decent amount of scale.
Yeah.
From a data.
That's awesome.
I was like, oh, I didn't, I think,
I feel like I, I feel like I
buried the, buried the lead there.
We are, we are a big group.
We take care of a lot of folks.
We're putting out meaningful papers.
Like this is why I feel like we
have kind of the raw inputs to, um,
to do this safely and effectively.
Which Martin comes to your questions.
It, it is a, it's more about
the safe and effective rollout
of this technology because when
you actually look at the need.
We were in West Virginia, for
example, 92% of the population
is on government healthcare.
But you have folks in these rural, very
rural areas that just do not have access.
They actually closed down a series of
OB GYN clinics and, and, uh, expecting
moms have anywhere from three to four
hour drive to get to, um, uh, to get to
a hospital to actually give birth, right?
These are, these are just
like fundamentally challenging
problems with access.
So I think AI has to play a role and it's
getting exponentially smarter, right?
You see what the, what the
Frontier Labs are doing.
I don't think you just jump
into prescribing, right?
So how I process this, and this
is actually pair team's approach,
is we're first starting with
navigation and orchestration.
So Flora, for example, can say, okay, I'm
interacting with you over voice or text.
Um, if I need to invoke an np,
I can schedule that resource.
If I need to coordinate with the
pcp, I can schedule that resource.
If I need to score, coordinate
with a housing organization, I can
coordinate that resource, right?
And over time, what you're gonna be
able to see, and we're already seeing
it now, what are the things that Flora
and these frontier models are better at?
And what are they technologically ready?
To do both, both I guess,
three components, which is also
regulatory compliance, right?
So that's the other ga
uh, rate limiting factor.
Um, but you ha by starting with
orchestration, you can see like,
well in this case it might actually
be better for me to use Flora to
prescribe in these certain XXY, Z cases.
And you can ab test it and you
can start to roll it out in
a much more data-driven way.
And you're looking at evals on
safety and also quality, uh, like the
experience that an individual gets.
Um, so we are optimistic, like over time.
I do think this, a lot of virtual care
is just gonna get taken over by agentic
tools and agentic com companionship.
Um, I think the question is going to be,
and it's gonna fill about this, this very
big need of access, but if you start with
orchestration and navigation, you can then
figure out safely where to push AI into,
um, to deliver the highest quality care.
Kevin: Neil on Access Model, not access
to care, but cmmis new access model.
The rolling out, yeah.
Big headline has been, there's been
a lot of consternation in digital
health community over rates, how folks
have been thinking about rates, and
we saw when the model was announced,
a lot of folks express their interest
in participating and then fallout
when, when rates, um, actually came
out and didn't apply to the program.
I, I'd be curious to hear your perspective
on how you're thinking about the
opportunity in the program and then how
you think about building the p and l
within the construct of that program.
CM Maya has been pretty clear in
saying like, you can't think about
it from the incumbent mindset.
You gotta build from the ground up.
How are you processing all of
that, interpreting it for your
organization as you're leaning in here?
Neil: Yeah.
Yeah.
So I'll start with, um, the,
we were expecting these rights.
Um, we were, we were not expecting
anything, anything higher.
Um, it's a feature,
not a bug, so to speak.
Right.
Uh, to, to drive down, you know, and
actually get to deflationary outcomes.
Right.
Um, so that was all,
that was all expected.
Um, I, what's really interesting is when,
you know, I talked to CMI and Abe about,
you know, how these rates were formed.
Um, it's like, oh, are you forming them?
Like, give us the methodology, right?
And instead of looking at it from a
value delivered perspective, the rates
are actually composed more bottoms
up from a cost to cost to deliver.
Perspective, right?
Which is very different, you
know, than how you'd develop a
value-based contract or anything.
It's like, well, we believe it's
gonna be cost X and let's do a cost
plus model on top of it, how we are
thinking about this internally is we're
looking at, 'cause we've have just
millions of data points on the actual
workflows and orchestration of care.
And we're saying, well over the course of
the next, well, is it, is it at what, what
percentage of automation can you do now?
What is going to come over
the next three to six months?
What's gonna come over the next year?
And you actually build this
kind of, um, you know, waterfall
roadmap for task automation there.
Things like scheduling with PCPs, right?
That no longer has to be something
that a care manager has to do.
She can, you know, he or she can
focus on higher leverage tasks
that then leads to our cogs there.
And we have this very kind of confident
roadmap in front of us on, um, on
making all this, all this work in with
an ag agentic, an agentic service.
Um, I think maybe to make it a little
bit more interesting for the, for the,
the listeners, uh, two, the big key
PO components are gonna be CAC and
they're going to be, um, like how do you
actually acquire folks at, at all this?
And then, um, I've had some
consternation around device cost and
blood pressure device costs there.
Um, and uh, on the.
The CAC side, I think there's
like, you know, this is really
incentivizing partnerships.
It's really incentivizing access,
uh, providers to go out, partner
with ACOs, partner with PCPs, right?
Um, so that you're not like, like D two
C online media spend is not gonna work.
You're not gonna be able to
pay for Google ads and Facebook
ads to make this program work.
You have to go through, um,
either build a great brand or
get channel partners in place.
Um, and then on, you know, so that's
one side, one side of the equation.
Um, and then the other is, um,
like things like device costs.
They're a hit now, but, uh, you know,
have talked to, talked to Jacob and,
and other folks in the, of my team.
There's like, these hardware
costs are going down.
I think they're, they're, there's
a, a thought that they're gonna
build able a tool directory.
So preferred vendors for
these devices, right?
Which will give, um, having, having the
US healthcare system be a buyer gives
you great discounts and he pass down
those discounts to access providers.
Um, so while it's a little bit of a pain
right now, I think we'll, we'll get over
it with the, with the support of CMI.
Martin: Neil, this has been great.
Where can folks get in touch if
they would like to learn more, wanna
chat our state Medicaid director
and wanna, wanna, wanna team up?
Neil: Yeah.
Yeah.
If you are any or any of the above,
if you're, you know, state Medicaid
directors health plans, we are taking
this program and we are actually
working to bring it into managed care.
So what does AI care management look
like and our flavor of it, right?
As whole person, it's the
social in the medical side.
Uh, please reach out to, to
myself N eil@pairteam.com.
Um, and, uh, if you're a builder in
the space, I think we have a lot of
the raw materials to build an AI care
manager in a way that is just higher
quality than anyone else can right now.
And we're pushing on it.
So if that excites you, um,
please, please reach out.
We'd love to, we'd love to talk.
Martin: Neil, thanks so
much for your time today.
We'll catch up soon.
Kevin: Good seeing you,
Neil: Neil.
Martin: Good luck
Neil: on the
Martin: move.
Neil: Alright you guys.
Bye.
Thank you.