May 27, 2025 • 22 mins
Interested in being a guest? Email us at admin@evankirstel.com
Ever wonder why nobody picks up your calls or reads your emails anymore? In this eye-opening conversation, Aaron Christopher "AC" Evans, founder of DRIPS, reveals why traditional outreach methods are increasingly ineffective in our distraction-filled world.
AC explains how our evolution into an "on-demand society" has fundamentally changed consumer psychology. When everyone is constantly engaged with content they actively choose – from social media to streaming services – traditional push messaging tactics that demand immediate attention are doomed to fail. The stark reality: about 80% of people simply won't answer calls from unknown numbers anymore.
DRIPS has pioneered a different approach through asynchronous, humanized AI conversations. Rather than forcing consumers to stop what they're doing, this technology allows meaningful interactions to unfold over days or even weeks, respecting people's time while still accomplishing business objectives. AC walks us through how their system differs from standard chatbots by interpreting natural language with sophisticated AI while maintaining tight compliance with complex regulatory requirements across healthcare and insurance industries.
The results speak for themselves. One national health insurance provider improved their star ratings from 3 to 4 across all contracts through DRIPS' technology. Another saw a tenfold increase in health risk assessment responses. For businesses struggling with diminishing returns from traditional outreach, AC offers practical advice on getting started – from simple text message "priming" before phone calls to implementing more sophisticated conversational AI. The key insight: when you give consumers a voice and meet them where they are, engagement dramatically improves.
Ready to stop annoying your customers and start having meaningful conversations that drive results? This episode reveals how the future of customer engagement is evolving beyond interruption marketing into thoughtful, AI-powered dialogue that respects consumers while delivering exceptional business outcomes.
More at https://linktr.ee/EvanKirstel
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:01):
Hey everyone, Fascinating conversation today,
as we talk about why yourmarketing and sales outreach
suck and how AI-drivenconversations can really be a
game changer.
With DRIPS AC, how are you?
I'm good.
Speaker 2 (00:15):
Evan, how are you doing?
Speaker 1 (00:16):
I'm doing well.
Thanks so much for joining.
That's one of your taglines.
Is your outreach sucks.
Pretty controversial and boldstatement.
Before that, maybe introduceyourself and DRIPS what's the
big idea these days?
Speaker 2 (00:32):
Yeah, myself, aaron Christopher Evans.
A lot of people call me AC.
I'm a father, I'm a startupfounder DRIPS my business now
I've been in it for about 11years started in 2014, really
kind of pioneered the whole ideaof conversational outreach.
So you know, texting membersand consumers.
(00:52):
Essentially, what we're doingtoday is we're working with the
biggest companies in the worldto reach out to their members,
to reach out to their prospects.
All the big I shouldn't say alla lot of the big healthcare
payer, health insurancecompanies and the majority of
the PNC insurance companies aswell are working with DRIPS for
their outreach efforts.
Speaker 1 (01:14):
Fantastic.
And why do you thinktraditional outreach is broken?
I think many of us have our ownpersonal experiences, but
what's the big picture?
Speaker 2 (01:23):
Yeah, you know, thinking all the way back, right
, I think it starts with, youknow, we have now become an
on-demand society, right?
So, uh, I talked about thisyears and years ago.
Uh, when, you know, shopify andSpotify and Amazon and Netflix
kind of became the norm.
Right, like we all of a suddenare getting used to being able
to pause live norm.
Right, like we all of a suddenare getting used to being able
(01:46):
to pause live television right,get same day delivery stream
anything we want.
Right, have unlimited, you know, content in our fingertips.
So, long story short, no one isbored anymore.
Right Back when we were bored,right Back when you and I were
growing up, you know the phoneringing was an exciting moment.
(02:07):
Right, it's like, oh my God,who is?
Who is calling me?
Speaker 1 (02:10):
Right, because, when you got an email, we got an
email it was like a big deal.
Everyone opened their emailback in the day, right.
Speaker 2 (02:20):
but now we are in about the opposite of that right
, where everybody is busy 100 ofthe time doing whatever it is
that they want to do.
Right doesn't necessarily meanthey're productive.
Busy just means they're busy,right?
They'd be busy scrolling oninstagram, and you bet they'd
rather do that than to get aphone call about, you know,
doing a policy review with theirauto insurance carrier.
(02:42):
So that's what kind of shifted.
You know psychology of people,right?
Like?
Commercials don't work, radiodoesn't work, right, phone calls
aren't being picked up, directmail is not being read, email is
not being read.
So all these things have onething in common, and is that
they are trying to steal yourattention.
They're trying to stop you fromdoing what it is that you'd
(03:05):
rather be doing and have youtake some action, right?
So if I call you, what are youroptions?
Right, you have to stop whatyou're doing and deal with this
phone call, or you ignore thephone call.
And you know, long story short,people are ignoring that type
of outreach, they're ignoringthat type of push messaging that
is trying to take their time.
Speaker 1 (03:26):
Well said.
So, Drips, you're talking a lotabout humanized AI, because
we're all AI, gen AI obsessed.
So what does that mean inpractice, and how is it
different from just your averagechatbot experience?
Speaker 2 (03:39):
Sure, yeah, again, the difference between what we
like to call conversations andpush messaging is we are
reaching out and holding anasynchronous conversation.
Right.
To hold a synchronousconversation, to get somebody on
the phone, they have to stopwhat they're doing.
Right, we can ask a question.
We can say hey, evan, we'refollowing up about your annual
(04:01):
wellness visit.
Have you had time to schedulethat yet?
Question mark you can respond.
Then.
You can respond later.
You can respond the next day.
If you don't respond within acertain amount of time, we can
bump you and push you and sayhey, evan, just making sure you
got my message right.
Should we jump on a call?
Can I help you?
You know, did you have anyquestions about this?
So I'm not taking your time.
(04:29):
Right, you can keep watchingNetflix.
You can keep working.
You can keep watching your kids.
You can do whatever you want.
So that's the trick, so tospeak, with conversations.
Right Is that it's anasynchronous thing.
So I'm giving people a voicethat they can respond when they
want to respond.
Right, and the interestingthing is they respond with all
kinds of stuff, right?
You all of a sudden learn thatyou know the person doesn't want
to pick up the medicationbecause they're having an
allergic reaction or they can'tpay their bill because they got
laid off or they already pickedup the prescription and there's
(04:53):
something wrong with the data.
So by opening up and askingquestions, you give these
audiences a voice that theydidn't have before, right?
Because if you just keepcalling them and they don't
respond, you're not hearing whatthe problem really is, right.
So that's kind of the trick isto opening up conversations,
listening people, giving them avoice.
(05:14):
The hard part is it's hard toknow what they're saying, right.
So we've been doing NLP andmachine learning modeling for
many years now.
It's definitely easier now withLLM, chatgpt, openai et cetera.
We can get better accuracy.
But the problem with generativefor most companies is that it
(05:36):
is probabilistic in nature,which means it's probably going
to do what you want it to do.
It doesn't mean it's going todo what you want it to do, so we
actually stay in a fulldeterministic model.
Currently we do use, you know,nlp and LLMs to do intent
mapping to figure out what theperson said.
So they say I'm driving, I'mstuck in traffic on I-77 South
(05:57):
heading into the office.
We know that doesn't mean I'mat work, it means I'm driving,
so that we would then send adeterministic, pre-prescribed
response that says oh sorry, wecaught you driving.
Please drive safe.
Let us know when you get towhere you're going, right?
So the trick is, you knowfiguring out what they're saying
so you can respond in kind.
That way you don't have tosound like a chatbot, right?
(06:20):
So the difference is withchatbots, it's uh, hey, evan, uh
, it's time for your annualwellness visit.
Respond c if you would like acall to confirm to blah blah
blah.
Respond r if you need toreschedule it.
And you know what a human doeswhen they feel like they're
being talked to by a robot.
They ignore it.
Right, they're like yeah I'll dothis later, you know what I
mean.
Like it's the equivalent ofgetting a phone call or saying
(06:42):
like hello, this is Walgreens, Iwon't use name, this is your
pharmacy, right?
It's not as empathetic, right?
So, but if a person calls me,or something that sounds like a
person calls me or texts me, I'mgoing to have a little more
empathy to hold a conversation.
Speaker 1 (06:59):
Well said.
Yeah, I don't even answer myphone anymore, so there goes
that.
Don't even answer my phoneanymore, so there goes that.
But why do you think so manycompanies, even big companies,
big brands, with deep pocketsand lots of tech in-house, are
so slow to evolve in this space?
They're still using emailblasts and text blasts and
outbound call centers.
Why haven't they jumped on theGen AI train?
Speaker 2 (07:22):
It's a good question.
Again, with Gen AI, there's alot of risks, right, and there's
a lot of uncertainty right now.
So most big companies justdon't want their data touching
those models.
Yet, right, most largecompanies, they are, you know, I
would say, implementing AI.
I was just with the CIOs of alarge, one of the largest
(07:44):
national health insurancecompanies.
I asked them how they'reimplementing Gen AI and they
said they're actually focused onimplementing it in-house,
meaning for operations, right,like getting it so that their
teams are using these tools sothat their company can be more
safer and more obvious place tostart with Gen AI than is using
(08:06):
it to outreach to your audienceon regulated channels.
So email is regulated, but it'srelatively pretty safe.
Direct mail you pay for a stamp, you're covered under the law.
Telephony calls and texts aregoverned by a litany of
(08:28):
different laws.
You have CMS laws for Medicare,medicaid.
You have TCPA, which is theTelephone Consumer Protection
Act Under the FCC.
You have the telemarketingsales rule Under the FTC.
You have the campaign register,tcr.
You have HHS for healthcaretype things.
You have different state levelrules, right, for telemarketing
(08:48):
and for collection.
So there's a really nastypatchwork of rules, regulations,
laws, best practices that acompany has to navigate to do
texts or calls, and what happensis, because it is so confusing
and conflicting, most legalteams relegate their business
(09:08):
owners to doing the lowestcommon denominator, which is a
phone call, right, or maybe apush text message.
That's not very engagingbecause they know that you know
one text one day, whatever it is, they can get it.
So it follows all the laws.
Holding a long tailconversation and doing text and
calling, and you know,responding to people.
(09:29):
It's just hard to do whilefollowing all the rules and
regulations.
And I'll give you a practicalexample.
So bill pay reminders right,like to do a bill pay reminder
saying that a bill is coming up.
You're allowed to mention theamount of the bill, right,
because it's not a collectionyet, it's just a bill that's
(09:49):
coming up the moment that thatbill is due the very next day.
In many states I think it'slike 23 or 24 states you're not
allowed to do what's calledthird-party disclosure, which is
essentially mentioning thatthere's a debt or a bill due,
right?
So you would have to say like,hey, we need to talk about your
account, right.
(10:09):
In other states you're allowedto mention the bill, right, it's
not a problem.
In Massachusetts, you're onlyallowed to reach out three times
within one week about somethingthat could be considered a debt
collection.
In a lot of states they followthe 777 collection rules, so you
can only reach out seven timeswithin seven days.
So there's, just like all thesedifferent rules, federal state
(10:32):
that you have to be able tofollow, depending on the use
case right.
Speaker 1 (10:36):
Is it?
Speaker 2 (10:37):
collections.
Is it healthcare related?
Is it marketing?
Is it administrativeinformation?
On nature and depending on thelevel of consent that you have,
was it an inquiry?
Do you have permission tocontact?
Have you been given priorexpress permission?
Have you been given priorexpress written consent?
It's just, it's a terriblething to try to organize around
(10:58):
In DRIPS.
We've actually built it intowhat we call the DRAE or the
DRIPS rules engine, so we have asystem that's essentially
policy by code to make sure thateverything that's coming out of
our system on behalf of ourclients is compliant at every
level.
Right, there are policies state, federal and all those
different rules, so differentmembers will have a different
experience based on where theyare, based on what the consent
(11:23):
level is.
So the short answer is it's hard, right.
I mean that's like in anutshell it's hard to do it well
in a deterministic fashionwhile following all the rules.
So therefore, we stick towhat's safe right.
We stick to email, direct mail,phone calls and you know the
diminishing returns are hittingright.
Like people, like you said,you're not picking up your call
(11:43):
at all.
That's a lot of people.
I think that's like we did astudy recently.
I think like 80% of people orsomething like that, just won't
pick up unknown callers.
So times are changing.
People are getting around tothis as a methodology.
They're seeing the lift fromjust doing push text message.
So they're trying to do betterby, you know, turning that into
a conversation.
There's a lot of different waysto do it but we feel like ours
(12:08):
is superior, fantastic.
Speaker 1 (12:09):
Let's talk the other hard thing, which is scale.
I get a text message from mydentist and I think it's
actually a human being.
They're actually just talkingto me and I'm not sure it's AI,
but either way, that works finefor a dentist's office.
But you get into millions ofusers.
That changes the game.
How do you scale to that sortof magnitude?
Speaker 2 (12:31):
You just nailed it.
I mean it probably is a humanright.
So there's, that's one of theways to do.
It is one-to-one, uh, causethen you're not regulated the
same way.
There's a lot less rules whenyou're hand texting somebody, uh
, and that does work for SMBs,right?
That's why we don't sell SMBscandidly, uh.
But yes, you're right at scale.
(12:52):
If it's a high value, highvolume problem, there are not
many ways to do it.
Well, right, there's dialers,uh, there, there's drips there,
there's companies that you knowwould like to compete with us
that do more.
One way, um, but I think oneway, and when I say one way, I
mean it's like it's either achatbot type thing or it's a
push excuse, excuse me, a pushalert.
(13:14):
Those are really good for highvolume, low value, right.
So if you think of a quadrant,right, you got low volume, low
value, which is like web chat,inbound support.
You have a high volume, highvalue, which is what we're
talking about, which isessentially usually us, or a
call center.
You know somebody that isstaffed out because these are
high value people and it's a bigvolume.
(13:34):
So that's why there's, you knowsomebody that is staffed out
because these are high valuepeople and it's a big volume.
So that's why there's, you know, a multi-billion dollar
industry around BPOs right.
Then you have the low value,high volume, which is alerts,
notifications right.
So you know, let's use a goodexample a gate change, right.
(13:54):
Like gates change all over theall over the place.
United is probably sending tensof thousands of these things
out a day, but it's not a highvalue, uh use case.
Because, evan, if you get atext saying that your gate
changed, guess what you're goingto do?
You're going to get your buttup and you're going to move to
the gate that you need to moveto right.
(14:15):
When it's high value, that'sbecause it's something that
enterprise needs right.
It's something that we need youto get a comprehensive medical
review.
We need you to do your Medicaidredetermination.
We need you to do a health riskassessment.
These are things that membersor prospects wouldn't do
(14:35):
otherwise or wouldn't do asquickly as the enterprise would
like them to, had it not beenfor the call center reaching out
, had it not been for DRIPSreaching out.
Right, if it's something thecustomer wants to do or needs to
do right finish an insuranceclaim, do a refund, dispute
something they're going to do itRight, there's many, many, many
(14:55):
softwares out there thousandsof different chatbots and web
support type softwares but whenit's something that the
enterprise wants the consumer todo, that's much more difficult,
right, and it's a lot harder toconvince them to take those
actions.
Speaker 1 (15:11):
Got it?
And what is the role for humanin the loop in best of class AI
and automation?
Different philosophies there.
What is yours?
Speaker 2 (15:22):
I think you know we have a really, really high
confidence score.
I think it's something like 99or 98.
I should know that when.
If it's below that meaning oursystem, our NLP and modeling
score is not 99 or 100% surethat we know the intent of the
user.
(15:42):
We send it to a human to review, to review, they essentially
assign what the correct intentshould be.
That then re-informs the modelright.
So many, many years ago it wasprobably 60, 40 humans.
Now, out of all the outreachour system does, I think less
than a percent or two is a human.
That's essentially assigningand training the model.
(16:04):
Because we've done so many ofthese right.
We've done billions andbillions of text messages.
So we've we've just candidlykind of seen all the different
rebuttals and different wayspeople will say things but,
human in the loop is reallyimportant.
I mean, like you know, you thinkof that doctor's office, right.
So maybe they have an alertthat goes out when your you know
(16:26):
time for your cleaning iscoming up every every months,
right, but when you respond back, because they probably don't
have really strong AI or NLPtypes of systems, maybe that's
when a human in the loop isthere, when they get a secretary
responding.
Problem at big scale is you'rejust shifting a call center from
(16:47):
being on the phone to a callcenter being on keyboards, right
, and that's still a reallycostly thing.
So certain things you knowaren't ready for our types of
automation yet.
You know, like really reallynuanced conversations,
conversations about, you know,medical type issues, that needs
to be done through a securechannel.
So we don't, we don't do thatover text message.
(17:07):
But you know, scheduling things, reminding people, getting them
to show up to things, makingsure they understand what's
important helping them alongtheir health care journey is.
We've pretty much cracked thatcode.
Speaker 1 (17:21):
Brilliant Speaking of cracking code.
You know there's no textbookabout best practices here.
There's no textbook about bestpractices here.
What advice do you give a CMO,maybe a CIO, who knows their
outreach sucks but doesn't knowhow to get started?
Speaker 2 (17:35):
What is a way to get started for wherever they are in
the journey.
That's a great question.
My favorite way to get started,like, a lot of companies are
doing great with calls already,right, a lot of companies are
doing some texting, right.
So I like to tell people to tryto marry those two things.
So they use texting generallyfor alerts, right?
(17:58):
Like, hey, we need you to do athing.
They treat it as a directresponse, which, again, I think
is not a good use of the channel.
I think you need to holdconversations to asynchronously,
take somebody down a journeyfor days or weeks.
But if you don't have that typeof technology, you don't have a
tool like DRIPS at yourdisposal, then why not marry the
awareness that a text bringswith the efficiencies and scale
(18:22):
that a call center brings?
And the way you do that is bypriming.
So you send a text messagefirst, right?
Hey, evan, this is yourhealthcare provider.
We're going to be calling youhere in a minute to talk about
scheduling your annual wellnessvisit.
It's important for X, y, z,right?
That alone.
That one text message alone, Iwould almost guarantee, could,
(18:44):
you know, increase your pickuprate or your contact rate by 20,
30, sometimes even 50%.
Right, because at least now theperson is aware and it and you
can play with that, right?
You can say, hey, we're goingto try to call you in 15 minutes
, right?
Or 10 minutes.
So now the person gets to maybestop doing what they're doing.
You know, put the kid away,take it, take a walk from the
(19:05):
office, whatever it may be, andtake your phone call right when
before, if it was just a phonecall, they're not picking it up,
right.
So that's an easy way to getstarted, you know.
But you know there are, look,there are partners and platforms
out there that specialize inthis.
You know that aresingle-focused, best-in-breed,
like DRIPS.
I think the best way to getstarted is look at what others
(19:26):
are doing, you know, look at thecase studies and test these
types of technologies.
Like I said, gen AI isn't therewhere big companies are, you
know, ready to trust it.
You know essentially, but thereare.
You know deterministic ways.
Even if you start with achatbot type technology, that's
(19:46):
better than nothing, right?
It's better than phone calls,in my opinion.
But if you can find aconversational outreach partner,
like DRIPS, or test out thattype of thing, I think that's a
great place to start.
Speaker 1 (19:58):
Brilliant, and for your more advanced practitioners
, customers of yours, they mustbe seeing some pretty incredible
ROI in terms of revenueattrition retention.
Do you have any anecdotes orstories there?
What do you typically see on?
Speaker 2 (20:15):
the investment they're making.
It's been great.
I mean we have a lot of casestudy.
It's been really exciting.
I'm trying to think of some,one of the most, a recent one
that was really exciting andwill be important in a couple of
years.
So we do a lot of work aroundstars.
So this is how these plans arerated right, and they have many,
(20:37):
many different measures.
Right.
How many people are doinghealth risk assessments?
How many people are doingconference medical reviews?
How many people are doingmedication?
How many people are medicationadherent?
Right, they're picking up andtaking their medicine.
We were able to increase onegroup's one of the big national
payers they're stars by, Ibelieve, a full point in all
(20:58):
their contracts.
Speaker 1 (20:59):
Essentially all the states.
Speaker 2 (21:00):
They moved from a three to a four point just
because of our work incomprehensive medication reviews
, comprehensive medic uhmedication reviews.
We also have started doing uhdigital assets.
We're building out essentiallyendpoints to help uh do surveys
for these payers, theseinsurance companies.
So we do health riskassessments now.
So we were able to show Ibelieve it was a 10x uh response
(21:23):
on health risk assessment.
So instead of getting a fewpercent, they're getting you
know upwards of 30 percent ofpeople and these are hard to
reach.
These are the people that don'tpick up the phone, don't do the
direct mail, don't get theemail.
So it's been very compelling.
I really do believe that whenyou give people a voice and
you're able to understand whatthey're saying back and you're
able to pivot and adjust theconversation to meet them where
(21:47):
they're at, while stillnurturing them for days or weeks
, like I said, the averageconversation takes more than a
week, right?
So this is not a direct response.
Hammer them till they pick up.
It's a very nuanced, you knowrelationship that you build with
these members so that you canbe empathetic and so that you
can really understand, you knowwhere they're coming from, while
helping them get to where youneed them to be.
Speaker 1 (22:10):
Wonderful Well tech for good and helping a much
underserved area of the consumer.
Well congratulations on all thesuccess.
Ac Thanks for joining andsharing the vision.
Speaker 2 (22:21):
Yeah, thanks for having me, evan.
I appreciate it.
Speaker 1 (22:23):
And thanks everyone for listening and watching.
Check out our new TV show yeah,broadcast TV.
That is this techimpacttvstarting this Saturday.
Thanks,
Hey everyone, Fascinating conversation today,
as we talk about why yourmarketing and sales outreach
suck and how AI-drivenconversations can really be a
game changer.
With DRIPS AC, how are you?
I'm good.
Speaker 2 (00:15):
Evan, how are you doing?
Speaker 1 (00:16):
I'm doing well.
Thanks so much for joining.
That's one of your taglines.
Is your outreach sucks.
Pretty controversial and boldstatement.
Before that, maybe introduceyourself and DRIPS what's the
big idea these days?
Speaker 2 (00:32):
Yeah, myself, aaron Christopher Evans.
A lot of people call me AC.
I'm a father, I'm a startupfounder DRIPS my business now
I've been in it for about 11years started in 2014, really
kind of pioneered the whole ideaof conversational outreach.
So you know, texting membersand consumers.
(00:52):
Essentially, what we're doingtoday is we're working with the
biggest companies in the worldto reach out to their members,
to reach out to their prospects.
All the big I shouldn't say alla lot of the big healthcare
payer, health insurancecompanies and the majority of
the PNC insurance companies aswell are working with DRIPS for
their outreach efforts.
Speaker 1 (01:14):
Fantastic.
And why do you thinktraditional outreach is broken?
I think many of us have our ownpersonal experiences, but
what's the big picture?
Speaker 2 (01:23):
Yeah, you know, thinking all the way back, right
, I think it starts with, youknow, we have now become an
on-demand society, right?
So, uh, I talked about thisyears and years ago.
Uh, when, you know, shopify andSpotify and Amazon and Netflix
kind of became the norm.
Right, like we all of a suddenare getting used to being able
to pause live norm.
Right, like we all of a suddenare getting used to being able
(01:46):
to pause live television right,get same day delivery stream
anything we want.
Right, have unlimited, you know, content in our fingertips.
So, long story short, no one isbored anymore.
Right Back when we were bored,right Back when you and I were
growing up, you know the phoneringing was an exciting moment.
(02:07):
Right, it's like, oh my God,who is?
Who is calling me?
Speaker 1 (02:10):
Right, because, when you got an email, we got an
email it was like a big deal.
Everyone opened their emailback in the day, right.
Speaker 2 (02:20):
but now we are in about the opposite of that right
, where everybody is busy 100 ofthe time doing whatever it is
that they want to do.
Right doesn't necessarily meanthey're productive.
Busy just means they're busy,right?
They'd be busy scrolling oninstagram, and you bet they'd
rather do that than to get aphone call about, you know,
doing a policy review with theirauto insurance carrier.
(02:42):
So that's what kind of shifted.
You know psychology of people,right?
Like?
Commercials don't work, radiodoesn't work, right, phone calls
aren't being picked up, directmail is not being read, email is
not being read.
So all these things have onething in common, and is that
they are trying to steal yourattention.
They're trying to stop you fromdoing what it is that you'd
(03:05):
rather be doing and have youtake some action, right?
So if I call you, what are youroptions?
Right, you have to stop whatyou're doing and deal with this
phone call, or you ignore thephone call.
And you know, long story short,people are ignoring that type
of outreach, they're ignoringthat type of push messaging that
is trying to take their time.
Speaker 1 (03:26):
Well said.
So, Drips, you're talking a lotabout humanized AI, because
we're all AI, gen AI obsessed.
So what does that mean inpractice, and how is it
different from just your averagechatbot experience?
Speaker 2 (03:39):
Sure, yeah, again, the difference between what we
like to call conversations andpush messaging is we are
reaching out and holding anasynchronous conversation.
Right.
To hold a synchronousconversation, to get somebody on
the phone, they have to stopwhat they're doing.
Right, we can ask a question.
We can say hey, evan, we'refollowing up about your annual
(04:01):
wellness visit.
Have you had time to schedulethat yet?
Question mark you can respond.
Then.
You can respond later.
You can respond the next day.
If you don't respond within acertain amount of time, we can
bump you and push you and sayhey, evan, just making sure you
got my message right.
Should we jump on a call?
Can I help you?
You know, did you have anyquestions about this?
So I'm not taking your time.
(04:29):
Right, you can keep watchingNetflix.
You can keep working.
You can keep watching your kids.
You can do whatever you want.
So that's the trick, so tospeak, with conversations.
Right Is that it's anasynchronous thing.
So I'm giving people a voicethat they can respond when they
want to respond.
Right, and the interestingthing is they respond with all
kinds of stuff, right?
You all of a sudden learn thatyou know the person doesn't want
to pick up the medicationbecause they're having an
allergic reaction or they can'tpay their bill because they got
laid off or they already pickedup the prescription and there's
(04:53):
something wrong with the data.
So by opening up and askingquestions, you give these
audiences a voice that theydidn't have before, right?
Because if you just keepcalling them and they don't
respond, you're not hearing whatthe problem really is, right.
So that's kind of the trick isto opening up conversations,
listening people, giving them avoice.
(05:14):
The hard part is it's hard toknow what they're saying, right.
So we've been doing NLP andmachine learning modeling for
many years now.
It's definitely easier now withLLM, chatgpt, openai et cetera.
We can get better accuracy.
But the problem with generativefor most companies is that it
(05:36):
is probabilistic in nature,which means it's probably going
to do what you want it to do.
It doesn't mean it's going todo what you want it to do, so we
actually stay in a fulldeterministic model.
Currently we do use, you know,nlp and LLMs to do intent
mapping to figure out what theperson said.
So they say I'm driving, I'mstuck in traffic on I-77 South
(05:57):
heading into the office.
We know that doesn't mean I'mat work, it means I'm driving,
so that we would then send adeterministic, pre-prescribed
response that says oh sorry, wecaught you driving.
Please drive safe.
Let us know when you get towhere you're going, right?
So the trick is, you knowfiguring out what they're saying
so you can respond in kind.
That way you don't have tosound like a chatbot, right?
(06:20):
So the difference is withchatbots, it's uh, hey, evan, uh
, it's time for your annualwellness visit.
Respond c if you would like acall to confirm to blah blah
blah.
Respond r if you need toreschedule it.
And you know what a human doeswhen they feel like they're
being talked to by a robot.
They ignore it.
Right, they're like yeah I'll dothis later, you know what I
mean.
Like it's the equivalent ofgetting a phone call or saying
(06:42):
like hello, this is Walgreens, Iwon't use name, this is your
pharmacy, right?
It's not as empathetic, right?
So, but if a person calls me,or something that sounds like a
person calls me or texts me, I'mgoing to have a little more
empathy to hold a conversation.
Speaker 1 (06:59):
Well said.
Yeah, I don't even answer myphone anymore, so there goes
that.
Don't even answer my phoneanymore, so there goes that.
But why do you think so manycompanies, even big companies,
big brands, with deep pocketsand lots of tech in-house, are
so slow to evolve in this space?
They're still using emailblasts and text blasts and
outbound call centers.
Why haven't they jumped on theGen AI train?
Speaker 2 (07:22):
It's a good question.
Again, with Gen AI, there's alot of risks, right, and there's
a lot of uncertainty right now.
So most big companies justdon't want their data touching
those models.
Yet, right, most largecompanies, they are, you know, I
would say, implementing AI.
I was just with the CIOs of alarge, one of the largest
(07:44):
national health insurancecompanies.
I asked them how they'reimplementing Gen AI and they
said they're actually focused onimplementing it in-house,
meaning for operations, right,like getting it so that their
teams are using these tools sothat their company can be more
safer and more obvious place tostart with Gen AI than is using
(08:06):
it to outreach to your audienceon regulated channels.
So email is regulated, but it'srelatively pretty safe.
Direct mail you pay for a stamp, you're covered under the law.
Telephony calls and texts aregoverned by a litany of
(08:28):
different laws.
You have CMS laws for Medicare,medicaid.
You have TCPA, which is theTelephone Consumer Protection
Act Under the FCC.
You have the telemarketingsales rule Under the FTC.
You have the campaign register,tcr.
You have HHS for healthcaretype things.
You have different state levelrules, right, for telemarketing
(08:48):
and for collection.
So there's a really nastypatchwork of rules, regulations,
laws, best practices that acompany has to navigate to do
texts or calls, and what happensis, because it is so confusing
and conflicting, most legalteams relegate their business
(09:08):
owners to doing the lowestcommon denominator, which is a
phone call, right, or maybe apush text message.
That's not very engagingbecause they know that you know
one text one day, whatever it is, they can get it.
So it follows all the laws.
Holding a long tailconversation and doing text and
calling, and you know,responding to people.
(09:29):
It's just hard to do whilefollowing all the rules and
regulations.
And I'll give you a practicalexample.
So bill pay reminders right,like to do a bill pay reminder
saying that a bill is coming up.
You're allowed to mention theamount of the bill, right,
because it's not a collectionyet, it's just a bill that's
(09:49):
coming up the moment that thatbill is due the very next day.
In many states I think it'slike 23 or 24 states you're not
allowed to do what's calledthird-party disclosure, which is
essentially mentioning thatthere's a debt or a bill due,
right?
So you would have to say like,hey, we need to talk about your
account, right.
(10:09):
In other states you're allowedto mention the bill, right, it's
not a problem.
In Massachusetts, you're onlyallowed to reach out three times
within one week about somethingthat could be considered a debt
collection.
In a lot of states they followthe 777 collection rules, so you
can only reach out seven timeswithin seven days.
So there's, just like all thesedifferent rules, federal state
(10:32):
that you have to be able tofollow, depending on the use
case right.
Speaker 1 (10:36):
Is it?
Speaker 2 (10:37):
collections.
Is it healthcare related?
Is it marketing?
Is it administrativeinformation?
On nature and depending on thelevel of consent that you have,
was it an inquiry?
Do you have permission tocontact?
Have you been given priorexpress permission?
Have you been given priorexpress written consent?
It's just, it's a terriblething to try to organize around
(10:58):
In DRIPS.
We've actually built it intowhat we call the DRAE or the
DRIPS rules engine, so we have asystem that's essentially
policy by code to make sure thateverything that's coming out of
our system on behalf of ourclients is compliant at every
level.
Right, there are policies state, federal and all those
different rules, so differentmembers will have a different
experience based on where theyare, based on what the consent
(11:23):
level is.
So the short answer is it's hard, right.
I mean that's like in anutshell it's hard to do it well
in a deterministic fashionwhile following all the rules.
So therefore, we stick towhat's safe right.
We stick to email, direct mail,phone calls and you know the
diminishing returns are hittingright.
Like people, like you said,you're not picking up your call
(11:43):
at all.
That's a lot of people.
I think that's like we did astudy recently.
I think like 80% of people orsomething like that, just won't
pick up unknown callers.
So times are changing.
People are getting around tothis as a methodology.
They're seeing the lift fromjust doing push text message.
So they're trying to do betterby, you know, turning that into
a conversation.
There's a lot of different waysto do it but we feel like ours
(12:08):
is superior, fantastic.
Speaker 1 (12:09):
Let's talk the other hard thing, which is scale.
I get a text message from mydentist and I think it's
actually a human being.
They're actually just talkingto me and I'm not sure it's AI,
but either way, that works finefor a dentist's office.
But you get into millions ofusers.
That changes the game.
How do you scale to that sortof magnitude?
Speaker 2 (12:31):
You just nailed it.
I mean it probably is a humanright.
So there's, that's one of theways to do.
It is one-to-one, uh, causethen you're not regulated the
same way.
There's a lot less rules whenyou're hand texting somebody, uh
, and that does work for SMBs,right?
That's why we don't sell SMBscandidly, uh.
But yes, you're right at scale.
(12:52):
If it's a high value, highvolume problem, there are not
many ways to do it.
Well, right, there's dialers,uh, there, there's drips there,
there's companies that you knowwould like to compete with us
that do more.
One way, um, but I think oneway, and when I say one way, I
mean it's like it's either achatbot type thing or it's a
push excuse, excuse me, a pushalert.
(13:14):
Those are really good for highvolume, low value, right.
So if you think of a quadrant,right, you got low volume, low
value, which is like web chat,inbound support.
You have a high volume, highvalue, which is what we're
talking about, which isessentially usually us, or a
call center.
You know somebody that isstaffed out because these are
high value people and it's a bigvolume.
(13:34):
So that's why there's, you knowsomebody that is staffed out
because these are high valuepeople and it's a big volume.
So that's why there's, you know, a multi-billion dollar
industry around BPOs right.
Then you have the low value,high volume, which is alerts,
notifications right.
So you know, let's use a goodexample a gate change, right.
(13:54):
Like gates change all over theall over the place.
United is probably sending tensof thousands of these things
out a day, but it's not a highvalue, uh use case.
Because, evan, if you get atext saying that your gate
changed, guess what you're goingto do?
You're going to get your buttup and you're going to move to
the gate that you need to moveto right.
(14:15):
When it's high value, that'sbecause it's something that
enterprise needs right.
It's something that we need youto get a comprehensive medical
review.
We need you to do your Medicaidredetermination.
We need you to do a health riskassessment.
These are things that membersor prospects wouldn't do
(14:35):
otherwise or wouldn't do asquickly as the enterprise would
like them to, had it not beenfor the call center reaching out
, had it not been for DRIPSreaching out.
Right, if it's something thecustomer wants to do or needs to
do right finish an insuranceclaim, do a refund, dispute
something they're going to do itRight, there's many, many, many
(14:55):
softwares out there thousandsof different chatbots and web
support type softwares but whenit's something that the
enterprise wants the consumer todo, that's much more difficult,
right, and it's a lot harder toconvince them to take those
actions.
Speaker 1 (15:11):
Got it?
And what is the role for humanin the loop in best of class AI
and automation?
Different philosophies there.
What is yours?
Speaker 2 (15:22):
I think you know we have a really, really high
confidence score.
I think it's something like 99or 98.
I should know that when.
If it's below that meaning oursystem, our NLP and modeling
score is not 99 or 100% surethat we know the intent of the
user.
(15:42):
We send it to a human to review, to review, they essentially
assign what the correct intentshould be.
That then re-informs the modelright.
So many, many years ago it wasprobably 60, 40 humans.
Now, out of all the outreachour system does, I think less
than a percent or two is a human.
That's essentially assigningand training the model.
(16:04):
Because we've done so many ofthese right.
We've done billions andbillions of text messages.
So we've we've just candidlykind of seen all the different
rebuttals and different wayspeople will say things but,
human in the loop is reallyimportant.
I mean, like you know, you thinkof that doctor's office, right.
So maybe they have an alertthat goes out when your you know
(16:26):
time for your cleaning iscoming up every every months,
right, but when you respond back, because they probably don't
have really strong AI or NLPtypes of systems, maybe that's
when a human in the loop isthere, when they get a secretary
responding.
Problem at big scale is you'rejust shifting a call center from
(16:47):
being on the phone to a callcenter being on keyboards, right
, and that's still a reallycostly thing.
So certain things you knowaren't ready for our types of
automation yet.
You know, like really reallynuanced conversations,
conversations about, you know,medical type issues, that needs
to be done through a securechannel.
So we don't, we don't do thatover text message.
(17:07):
But you know, scheduling things, reminding people, getting them
to show up to things, makingsure they understand what's
important helping them alongtheir health care journey is.
We've pretty much cracked thatcode.
Speaker 1 (17:21):
Brilliant Speaking of cracking code.
You know there's no textbookabout best practices here.
There's no textbook about bestpractices here.
What advice do you give a CMO,maybe a CIO, who knows their
outreach sucks but doesn't knowhow to get started?
Speaker 2 (17:35):
What is a way to get started for wherever they are in
the journey.
That's a great question.
My favorite way to get started,like, a lot of companies are
doing great with calls already,right, a lot of companies are
doing some texting, right.
So I like to tell people to tryto marry those two things.
So they use texting generallyfor alerts, right?
(17:58):
Like, hey, we need you to do athing.
They treat it as a directresponse, which, again, I think
is not a good use of the channel.
I think you need to holdconversations to asynchronously,
take somebody down a journeyfor days or weeks.
But if you don't have that typeof technology, you don't have a
tool like DRIPS at yourdisposal, then why not marry the
awareness that a text bringswith the efficiencies and scale
(18:22):
that a call center brings?
And the way you do that is bypriming.
So you send a text messagefirst, right?
Hey, evan, this is yourhealthcare provider.
We're going to be calling youhere in a minute to talk about
scheduling your annual wellnessvisit.
It's important for X, y, z,right?
That alone.
That one text message alone, Iwould almost guarantee, could,
(18:44):
you know, increase your pickuprate or your contact rate by 20,
30, sometimes even 50%.
Right, because at least now theperson is aware and it and you
can play with that, right?
You can say, hey, we're goingto try to call you in 15 minutes
, right?
Or 10 minutes.
So now the person gets to maybestop doing what they're doing.
You know, put the kid away,take it, take a walk from the
(19:05):
office, whatever it may be, andtake your phone call right when
before, if it was just a phonecall, they're not picking it up,
right.
So that's an easy way to getstarted, you know.
But you know there are, look,there are partners and platforms
out there that specialize inthis.
You know that aresingle-focused, best-in-breed,
like DRIPS.
I think the best way to getstarted is look at what others
(19:26):
are doing, you know, look at thecase studies and test these
types of technologies.
Like I said, gen AI isn't therewhere big companies are, you
know, ready to trust it.
You know essentially, but thereare.
You know deterministic ways.
Even if you start with achatbot type technology, that's
(19:46):
better than nothing, right?
It's better than phone calls,in my opinion.
But if you can find aconversational outreach partner,
like DRIPS, or test out thattype of thing, I think that's a
great place to start.
Speaker 1 (19:58):
Brilliant, and for your more advanced practitioners
, customers of yours, they mustbe seeing some pretty incredible
ROI in terms of revenueattrition retention.
Do you have any anecdotes orstories there?
What do you typically see on?
Speaker 2 (20:15):
the investment they're making.
It's been great.
I mean we have a lot of casestudy.
It's been really exciting.
I'm trying to think of some,one of the most, a recent one
that was really exciting andwill be important in a couple of
years.
So we do a lot of work aroundstars.
So this is how these plans arerated right, and they have many,
(20:37):
many different measures.
Right.
How many people are doinghealth risk assessments?
How many people are doingconference medical reviews?
How many people are doingmedication?
How many people are medicationadherent?
Right, they're picking up andtaking their medicine.
We were able to increase onegroup's one of the big national
payers they're stars by, Ibelieve, a full point in all
(20:58):
their contracts.
Speaker 1 (20:59):
Essentially all the states.
Speaker 2 (21:00):
They moved from a three to a four point just
because of our work incomprehensive medication reviews
, comprehensive medic uhmedication reviews.
We also have started doing uhdigital assets.
We're building out essentiallyendpoints to help uh do surveys
for these payers, theseinsurance companies.
So we do health riskassessments now.
So we were able to show Ibelieve it was a 10x uh response
(21:23):
on health risk assessment.
So instead of getting a fewpercent, they're getting you
know upwards of 30 percent ofpeople and these are hard to
reach.
These are the people that don'tpick up the phone, don't do the
direct mail, don't get theemail.
So it's been very compelling.
I really do believe that whenyou give people a voice and
you're able to understand whatthey're saying back and you're
able to pivot and adjust theconversation to meet them where
(21:47):
they're at, while stillnurturing them for days or weeks
, like I said, the averageconversation takes more than a
week, right?
So this is not a direct response.
Hammer them till they pick up.
It's a very nuanced, you knowrelationship that you build with
these members so that you canbe empathetic and so that you
can really understand, you knowwhere they're coming from, while
helping them get to where youneed them to be.
Speaker 1 (22:10):
Wonderful Well tech for good and helping a much
underserved area of the consumer.
Well congratulations on all thesuccess.
Ac Thanks for joining andsharing the vision.
Speaker 2 (22:21):
Yeah, thanks for having me, evan.
I appreciate it.
Speaker 1 (22:23):
And thanks everyone for listening and watching.
Check out our new TV show yeah,broadcast TV.
That is this techimpacttvstarting this Saturday.
Thanks,
Popular Podcasts
True Crime Tonight
If you eat, sleep, and breathe true crime, TRUE CRIME TONIGHT is serving up your nightly fix. Five nights a week, KT STUDIOS & iHEART RADIO invite listeners to pull up a seat for an unfiltered look at the biggest cases making headlines, celebrity scandals, and the trials everyone is watching. With a mix of expert analysis, hot takes, and listener call-ins, TRUE CRIME TONIGHT goes beyond the headlines to uncover the twists, turns, and unanswered questions that keep us all obsessed—because, at TRUE CRIME TONIGHT, there’s a seat for everyone. Whether breaking down crime scene forensics, scrutinizing serial killers, or debating the most binge-worthy true crime docs, True Crime Tonight is the fresh, fast-paced, and slightly addictive home for true crime lovers.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.