December 16, 2025 • 40 mins
Dan Balcauski talks with Ryan Wang, CEO and co-founder of Assembled, a support operations platform. The discussion covers Assembled's evolution from a workforce management tool to incorporating AI capabilities for customer support, the cultural and strategic challenges of launching new products, and lessons learned from experiences at Stripe. Ryan emphasizes the importance of a clear mission, customer feedback, and managing internal dynamics when expanding product offerings. They also explore the practicalities of introducing new technologies, team coordination, and building a cohesive company narrative.
01:46 What Assembled Does
02:38 The Evolution of Assembled
06:33 Building a New Product
11:37 Cultural and Strategic Lessons
20:30 Comparing Stripe and Assembled: Cultural and Systemic Differences
23:35 Launching New Products and the Role of AI
24:16 Tactical Approaches to Market Entry
26:02 Challenges in Scaling Sales Teams for New Products
29:55 Strategic Decisions on Product Structuring
Guest Links
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Ryan Wang (00:00):
Yeah it started as a second product and then we
(00:02):
accidentally turned into three,four, and five all at once.
in 2022, of course chat GTlaunched and.
With our ML backgrounds and justkind of surveying the technology
landscape it became reallyobvious to us, well, what is
possible in customer support hastotally changed and we should
not build AI for support becauseit's cool or because the
(00:24):
technology seems really fun toplay around with.
But it actually did start intruth with somebody at a
hackathon putting together a botand dropping into Slack.
When you talk to customers,magical things happen.
my belief is that, you wincustomers fundamentally by word
of mouth.
Awareness helps but at the endof the day, there's nothing that
beats finding somebody, havingthem take a bet on you.
Dan Balcauski (01:08):
Welcome to SaaS Scaling Secrets, the podcast
that brings you the insidestories from the leaders of the
best scale up.
B2B SaaS companies.
I'm your host, Dan Balcauski,founder of Product Tranquility.
Today I'm excited to welcomeRyan Wang, CEO and co-founder of
assembled a support operationsplatform that helps companies
like Stripe, zoom, and Etsymanage their customer support at
scale.
Before founding assembled, Ryanwas employee number 80 at
(01:28):
Stripe, where he watched thecompany scale from 80 to 800
people and helped build the MLsystems that powered fraud
detection and supportautomation.
Ryan, welcome to the show.
Ryan Wang (01:37):
Great to be.
Dan Balcauski (01:39):
Very excited for our conversation today.
Ryan, before we dive into yourscaling journey, just give us
the elevator pitch.
What does Assembled do?
Who do you serve?
Ryan Wang (01:46):
Yeah, assembled is the AI platform for customer
support.
So we help companies with thefull gamut of all things
customer service.
So automation to solve 70, 80%of your customer tickets over
chat, over voice, over emailaugmentation.
So co-pilot that makes people20, 30% more
Dan Balcauski (02:02):
I.
Ryan Wang (02:02):
ramp 50% faster.
All the way through tooperations.
So helping companies manage20,000 plus person contact
centers with the forecasting andthe staffing planning and the
capacity planning betweenultimately whether you should
talk to an AI agent or whetheryou should talk to a human agent
or whether that should beinterspersed, making sure your
customers have the rightexperience.
We help companies of alldifferent shapes and sizes, so
(02:23):
Ashley Furniture, Stripe, RobinHood.
So, so it's quite the complexset of customer service
concerns.
Dan Balcauski (02:32):
Well quite a broad portfolio and I wanna dive
into that set of capabilitiesdirectly.
So as I understand it, youcorrect me if I'm wrong, but
assembled started as a workforcemanagement.
Platform and now you've addedthese AI agent and copilot
capabilities.
And so adding that, maybe youconsider it a second product or
(02:52):
not, but it kind of appears thatway, could often be a
challenging transition forscaling companies.
So I'm really curious about thedecision around adding.
Additional product becauseagain, this could be a fraught
time.
Um, what was your thoughtprocess in terms of timing?
Like how did you think aboutthis is the right time to do it
(03:13):
or not?
And we go from there.
Ryan Wang (03:15):
Yeah it started as a second product and then we
accidentally turned into three,four, and five all at once.
And so really stretched ourroadmap.
The story goes further back.
So I was at Stripe.
I was actually working onmachine learning for fraud
detection.
This was around 2014 or so.
And had nothing to do withcustomer service, but Patrick
and John Collison, theco-founder of Stripe.
(03:37):
They cared about it a lot.
They would do support rotations.
They would have people over totheir apartment in the mission
in San Francisco and invite thewhole company to do customer
support.
And then over time, of course,from 80 to 800, that doesn't
make sense.
It doesn't scale.
You bring in outsourcers, youstart to have different products
that you're servicing acrossdifferent geographies.
(03:59):
You start to need to havedifferent tiers of escalation.
And so, it became really obviousright around 20 17, 20 18, that
customer support was one ofthose spaces where you would
apply, not ai.
But machine learning and it wasin service of helping the
company scale, great customersupport, being high quality,
(04:19):
really efficient multi-product,global support.
And so when we started, ourfirst product was workforce
management.
But that actually came out of aquestion that we asked to the
folks at Stripe.
There was a guy named Bob VanWinden who ran support and he
had come from Google and seenthe challenges scaling up
Google, and we asked him justWhat is software you need to buy
(04:42):
or you hit your options.
And it wasn't just Bob, it was ahundred other support leaders
but a lot of them came back withthis answer of, well.
This thing called workforcemanagement.
It's a team that's growing from,several hundred to several
thousand and figuring out whoshould be working on what,
getting the right person in theright place at the right time.
That was the challenge of 2018in terms of scaling support and
(05:02):
exacerbated in 2020 when welaunched.
So it actually took us two yearsto get off the blocks.
But in 2020, companies weregrowing really quickly, and it
was all through headcount.
And so workforce managementbecame an even bigger problem.
And then in 2022, of course chatGT launched and.
With our ML backgrounds and justkind of surveying the technology
landscape it became reallyobvious to us, well, what is
(05:23):
possible in customer support hastotally changed and we should
not build AI for support becauseit's cool or because the
technology seems really fun toplay around with.
But because that answer to thatquestion, how do you scale great
customer support.
We feel like has changed in thismoment.
So, it was really easy for us tosay, yeah, that's something we
(05:44):
wanna do.
It's something that fits ourmission.
It's something that fits ourproduct portfolio in terms of
the value we deliver and andthen stack these products
together.
Dan Balcauski (05:53):
So, so solving the same core problem and this
happenstance of the technologymeeting capabilities with your
background experience, uh,around, you know, the ability to
bring ML to really solve thisproblem well outside of just
adding humans to the mix.
So you guys decide to go fullbore on this.
(06:15):
How did you approach it?
Like meaning, how did you thinkabout organizing the team to
build this additional set ofcapabilities?
And maybe at that time it wasmaybe, I dunno if you thought of
it originally as a net newproduct or just a feature set,
but kinda lead me through thatprocess.
Ryan Wang (06:33):
Yeah we did think about it as a new product from
the very beginning, I will sayit was.
2.5, possibly take three ofbuilding a new product that
assembled and so company hadactually come out of this
bottoms up motion at Stripe.
We were building ML tools forcustomer support.
But I can tell you.
(06:54):
asked us to do it.
They actually, they asked us notto do it.
I remember having this meetingwith the head of product and the
head of systems engineering andthe head of product said, Hey,
you should be working on thedashboard.
That's a high priority.
There's not enough peopleworking on that.
Why are you working on support?
And the head of systemsengineering said, you're doing
machine learning for customersupport.
Doesn't NLP solve all this,kinda shades of AI doesn't like
(07:15):
what's the point?
Isn't this just kind of it onand it works?
So, so actually the firstproject within Stripe that was
kind of ML support happened.
It was almost a startup.
Within a startup.
It was.
myself and one of my co-foundersBrian Z trying to solve a
problem, going to visit callcenters, figuring out what their
problems were trying to pull inmore engineers and scrap for
(07:39):
resources.
We borrowed this intern, said,Hey, we have this project going
on and it's starting to delivera lot of impact and we think it
could be a really big teameventually.
So that was our mindset cominginto assembled of, okay, well we
had created a product and a lotand that wasn't just customer
support.
A lot of the great products thatStripe had come bottoms up
somebody.
There's this thing called theCrazy Ideas Hackpad.
(08:02):
looked on the crazy ideasHackpad, and they were hired to
do this, but they said, actuallyit would be cool to have an API
to create a company.
That was how Shred Ball cameabout.
So our first.
product.
We just thought, hey, if we tellpeople, you know what, here's
the story of how assembled evengot started and if you see the
right opportunity, you don'thave to ask, just do it.
(08:23):
Create a new product.
And it never off the got off theground.
There was a group of people thatwould meet weekly and they were
trying to build a product calledRoute Automated Routing, and now
it's part of our productportfolio.
But they would meet every weekand it just never caught.
And then our second attempt at aproduct.
It turned out to be a killerfeature.
It was a feature to help BPOs oroutsourcers connect into
(08:46):
customers like many of ourlargest customers.
So, so if you have multipleBPOs, it's almost like a network
of.
Rather than individuals,hundreds of individuals at a
time.
So we wanted to give peoplevisibility in that whole network
rather than just the people.
And we came super tops down.
We wrote a strategy.
A woman who is an MBA internwrote a really impressive
(09:09):
document on all the BPOs weshould go after in what order,
the integrations we would needthe size of the market.
She's now at OpenAI.
So it was really high quality,but just took us so long to get
off the blocks there.
And then the minute we startedbuilding it was like, hmm, the
strategy's not quite right.
Actually, we gotta start withthe integrations, not the BPOs
this way.
So our new product around AI waskind of the synthesis of those
(09:31):
two things was.
We know strategically why we'redoing this.
We want to help deliver greatcustomer support.
We know the technology LMS aregonna change everything in terms
of what's possible.
But it actually did start intruth with somebody at a
hackathon putting together a botand dropping into Slack.
And we had this Slack channelcalled product Questions.
(09:55):
People come ask, Hey, how doesour in Workday integration work?
And in truth, it was, product,70% of the time you get an
answer to your question.
but when this bot showed up andit was automatically answering
questions, our support team, oursales team, our customer success
team, they were, oh my gosh,we're actually getting answers
to our questions, and they'reright and they're good.
(10:16):
So from there it became reallyobvious, okay, we have
something.
There's something to buildaround.
There's a use case.
And then the top down version oftaking that idea and running
with it was to say, we're gonnaput you in a separate room this
time around.
We're gonna put you in adifferent Slack group where
you're gonna, gonna have peoplededicated to this team and your
(10:36):
whole mission, and your wholegoal is to create a new product.
And we're gonna borrow theplaybook.
Even of Y Combinator, you have12 weeks.
At the end of 12 weeks, you'regonna do a demo day.
And.
The fast forward is that itdidn't take 12 weeks.
It took a little bit longer than12 weeks, but it operated in
that kind of cycle.
Dan Balcauski (10:54):
So just to, so I could lay out the roadmap for
listeners a little bit.
So, the AI product not thesecond product necessarily, but
you described it, maybe it's,2.5 or three, depending upon how
you look at it with thisinterim, products.
That was for the BPO.
So you have sort of workforcemanagement, uh, this BPO, uh,
initiative, and then you havethe new AI agent and copilot
(11:17):
offering.
Now I'm curious you, so you saidthe, you, so you separate this
group off.
Y Combinator style to go run for12 ish weeks.
What were you trying to avoidwith that situation?
I guess what was the lessonlearned from that intermediate
time where you're like, oh, thiswill be the better way to go?
Ryan Wang (11:37):
Yeah.
Yeah.
The lesson from the firstexperiment was that it needed to
be a dedicated team with a cleargoal.
Dan Balcauski (11:43):
Hmm.
Ryan Wang (11:43):
it couldn't just be purely bottoms up.
And so, for example.
We, we didn't tap people.
We didn't say, Hey, you're, oneof our strongest engineers.
Go join this team.
Instead.
We had an application process.
The application was just threeshort questions.
It wasn't super formal.
Have you done at assembled sofar?
(12:04):
why do you wanna join this team?
And what do you know about ai?
you could kind of, tease apartmost of the characteristics that
we need in the team from thosequestions.
But the lesson from the secondangle of getting too hung up in
strategy, not going too topsdown was just, it's the classic
Mike Tyson quote.
Everybody's got a plan untilthey're punched in the face.
(12:25):
We had this plan, we just don'tknow.
The space is evolving, thetechnology's evolving, and then
also in the early days ofassembled, it's not like we set
out and said.
We're going to go solveworkforce management.
what we said is we're going togo improve customer support.
That's the broad bucket in whichwe play.
And then we're gonna go ask abunch of people who know a lot
about customer support, what isthe specific problem?
(12:45):
So we wanted to put people intothat situation where.
could just ask that question,run after it, learn really
quickly, be unencumbered by aplan that was to set based on
things that, that they didn't,we didn't learn from customers.
And then have a clear goal atthe end of the day.
Dan Balcauski (13:02):
So, okay, so you put these folks off Tiger Team
Isolation dedicated process witha high maybe some sort of filter
into that group that, you know,these folks are qualified
go-getters, uh, that you want inthere.
So you know, they go run off.
I guess, what was your idea for,okay, like they're gonna go
create something.
(13:22):
How do you, was the idea like,okay we'll then be just kind of
creating like a, another companydivision.
Was it like, what was thethought about like how you're
going to reintegrate them, Iguess?
Tell me what that looked likewhen, what's that 12 ish week
period is done?
Ryan Wang (13:38):
Yeah.
Well this part we really messedup, I would say.
And when you read the.
HVR McKinsey articles aboutHorizon 1, 2, 3.
They make it sound so clean.
You're in the zero to one phaseand then you know when you're
out and then you put it in thisother stage.
It just.
Maybe we didn't read thearticles closely enough, but we
definitely messed up on thereintegration part.
(13:59):
I think because it took a reallylong time for it to be clear
that it would be successful.
In truth, at the end of the 12week demo day there was some
really exciting progress.
There was some, adoption chartsthat were going up, but it
wasn't this is in 2023.
It wasn't like, oh geez.
E everything sorted out and youcould see this thing going from
zero to a hundred million a rovernight.
(14:20):
It was, you kind of had tosquint a little bit
Dan Balcauski (14:23):
Mm.
Ryan Wang (14:23):
There was early customers where, okay, they're
really sticking with what we'redoing, but we have to keep
iterating here so.
The 12 week demo day turnedinto, you know what, we'll check
back in again in three months.
And then instead of it justbeing a demo day for a small
group of people, you're gonna goin front of the whole company.
It'll be at the all hands.
And so, you'll have to do thatpitch for everybody.
(14:46):
And even then, I think.
The mistake was we had kind ofset it up as the company is
almost like judging you,American Idol style it created a
little bit of resentment.
come these people who, filledout the application, they got to
be on the team and we're overhere.
Working on the old thing.
They get to mess around with AIand it's super fun and they were
(15:09):
having a lot of fun in the room.
I was like, how come they'rehaving so much fun?
This is what the heck, we'reover here solving big,
multimillion dollar a yearcustomer problems, and they're
over there just throwing stuffat the wall and moving fast.
The hell, so, so that resentmentlingered I think when it became
really clear, okay, this issuccessful.
We want more to put moreinvestment for us of the
(15:29):
company.
We wanna move people from ourolder product to our newer
product.
And we get this question all thetime.
Are we pivoting the company?
Are we pivoting the company andhow do these products go
together and should we jet us inthis off?
Or the small group said, maybewe should turn this into a
separate startup and it's asubsidiary
Dan Balcauski (15:46):
Mm.
Ryan Wang (15:46):
And in both directions.
And that resentment was reallyproblematic because the story we
tell about assembled that I justtold you is.
Well, our value prop is to behuman and ai.
It's to own the end-to-endcustomer experience.
And all the products on top ofthat, like these products are
stronger together and they'reall about helping our customers
deliver better customer support,high quality, super efficient.
You can't do that if it's justone piece of this kind of narrow
(16:09):
slice.
But we really, really had tobring people back into that
because of, I think some of thisresentment, some of this
cultural built up story.
And then also because we wentinto it without a plan.
Dan Balcauski (16:21):
Mm-hmm.
Ryan Wang (16:21):
without a plan, but without, but by saying, who
cares about the strategy?
We'll figure it out after it'ssuccessful.
But then once it was successful,we really had to build the story
back.
Dan Balcauski (16:29):
I, well, I'm curious.
Uh, it's funny because I'veactually, uh, my previous life
as a product leader, I got putin one of these labs teams as
like, you're gonna horri,basically Yeah.
These Horizon three group.
And so I've lived that.
World of resentment.
The irony is, is like it's notany better on the other side.
Like, like, like they're alwayshaving fun over there.
(16:52):
They caught you in a moment oflike gallows humor probably
because like you're over therebeating your head against the
wall, being like, it's still notworking.
It's still not working.
Ryan Wang (17:00):
Yeah.
Dan Balcauski (17:00):
Our time is running out budget is running
out.
Like we've got show progress.
I'm curious, like as a leaderyou know what, I guess you know,
was that.
People, people's feelings Iguess are always valid.
Do people have feelings?
But I'm sure like there was aperception issue.
Like are there things that, youfound that were successful in
diffusing that, or things thatguys looking back that you like
(17:22):
would have like implemented?
Like just think about otherpeople listening if they're
gonna like, send off these cyberteams.
Right.
Like it's interesting'cause youalways think about like, okay,
what's the market acceptance ofthis thing?
And I could easily see how thatsort of cultural resentment.
Maybe like it was like, oh, itwasn't even on the checklist of
things I needed to worry about,and all of a sudden I'm having
to manage that.
So like as you think about that,reflect on that time as a leader
now, like is there things thatyou did to help push that
(17:45):
forward or, even if it was laterthan you might have anticipated.
Ryan Wang (17:49):
Well, I resonate a lot with what you said.
It feels like outside lookingin, it's so fun.
And then inside looking out andthis was the resentment going
the other way that we saw too.
Like, what the heck?
This is really hard.
It's existential, it's like wedon't even have any customers
yet.
Or, Hey, we have like threecustomers.
We're trying to get to 20.
How come the rest of the companywon't help us?
And I think a lot of the re theliterature, there's so much
(18:09):
great literature now on how togo multi-product, the mechanics
of it, and when to do it and whyto do it earlier.
There's not much on thiscultural piece for me.
I think lesson one or the first,most useful thing, and I, it
wasn't necessarily that I wentin prepared.
I talked to this guy, PeterGasner, who started a company
called Veeva,$50 billion LifeScience SA Company.
(18:33):
And this was about a, this wasin 2022.
Or so so before we even startedthis initiative, and he had told
me in the early days of Viva,they built a second product and
intentionally he said, we pullit as far from our first one as
humanly possible.
Why?
Because the tendency is to govery iterative.
(18:53):
So in Viva, in the early days,second product as far away as
possible, from the first one hesaid, we almost broke the
company.
This is, we almost broke thecompany and he elude.
I didn't understand at the time,but he alluded to the reasons it
was, I think because of thecultural pieces.
It's these questions.
Yeah.
Are we pivoting?
We thought we were workforcemanagement.
No, we, this company like.
(19:14):
joined the company and thatproduct was working.
No, you joined the company.
I thought we were improvingcustomer support.
Full stop.
So you have to lift people's eyeline.
You have to show them, Hey,yeah.
Dan Balcauski (19:24):
Mm-hmm.
Ryan Wang (19:24):
today's world, in any given moment, there's
constraints.
is competing against that interms of resources.
Go here or there.
But in the long term, all thesethings are coming back together
to be a powerful platform.
And you really have to showpeople that way.
And I do think that's partstorytelling on the part of the
leadership.
That was, I would say the secondpart.
And that contributes to thecultural story of why are we
(19:46):
doing this?
Are we doing this because weneed more a RR?
Are we doing this because AI issuper hot?
Are we doing this?
Because the other thing's notworking.
No, we're doing this becausethese things are better
together.
But that's a big part of keepingpeople motivated and them
understanding the story andseeing it for themselves.
Even though, again, just becauseof the, how far apart these
things are, they might have tosquint a little bit and they
(20:08):
might not get it for a while.
Dan Balcauski (20:10):
Well, I mean, uh, is the old, uh, adage, nature of
bores a vacuum and humans, ifthere's no information, we tend
to paint in that black spot ourworst visions of it.
Right.
So what you mentioned is whatwe're doing not working?
Or is this like another bet thecompany thing to get us, you
know, you're like, no, no, no.
It's all, all in the same plan,but we've gotta experiment.
(20:30):
So I could see how that be achallenge.
You mentioned before, I wanna.
We'll come back to kind of,launching with this new product,
but I want to just tie it backbecause you did talk a little
bit about Stripe and how StripeAtlas came out of this, kind of,
sort of bottoms up and I guesswhat.
What do you feel like was,either different in the water of
what you saw at Stripe versuswhat you experienced at
(20:52):
assembled?
I mean, maybe you were just in adifferent position as a, you
know, if you're, you know, anengineer kind of working on
stuff versus now you're the CEleader.
Maybe, you know, maybe the Coonshad the exact same view of like,
dysfunction, right?
Of like, oh my God, there's abunch of people launching stuff
and nothing's connected.
I don't know, I haven't pickedtheir brain.
Um, be happy to talk to'em atsome point, but, um, I'm
curious, like, kind of lookingback like what do you, like, was
(21:13):
it a cultural thing?
Was it like part of the systemsand processes they had for kind
of taking those products fromincubation into the market?
Like what do you what was yoursense of like maybe what was
different?
Ryan Wang (21:24):
Yeah, I think the process and systems came later,
but I think two big dimensionsthat worked in concert, I think
one was that kind of clarity ofmission.
That was a really big one.
And Stripe, they're not apayments company.
It's increasing the GDP of theinternet.
(21:46):
And when I was an engineer, Ithought, that's so stupid.
What, what does that mean?
Increase the GDP of theinternet?
No.
What?
Like, let's build technology,let's improve payments.
I don't get it.
but now when you zoom though allthe way out, it's like, how does
Bridge Stablecoin fit in?
How does Stripe Atlas fit in?
How does having.
(22:06):
printing press, you know that.
Does books make sense?
Well, yeah.
It's not a payments company.
It's increasing the GP of theinternet.
All of these bets fit under thatbig tent.
And similar for us, I thinkwe're trying to improve customer
support, full stop.
But even beyond that, we wannasolve the operational challenges
that come after big ideas.
For companies like Stripe thatbecome successful.
Support is a function.
(22:27):
Trust and safety is a function.
Operations broadly is stuff thatcomes after.
So we try to fit people intothis.
It's not a workforce managementcompany, that's one of our
products for sure, but we'retrying to solve this very big
problem.
Dan Balcauski (22:39):
Hmm.
Ryan Wang (22:39):
I think the other part that fits that plays well
with a big vision is.
When, what the best people inthe world.
They don't wanna just, yeah, Idon't want, just wanna make a,
10% better payments company thanthe thing that came before.
Don't wanna just build customersupport systems and solve this,
legacy problem that's a littlebit better.
They wanna do big things.
And just to put that inperspective, my recruiter at
(23:02):
Stripe, Daniella Amedee, she'spresident of Anthropic my.
Onboarding was done by the thenCTO of Stripe.
Greg Brockman, he's of OpenAI,and so you know, you get this
density of people together whoare not just super, super smart,
super, super humble, all ofthose things, but just want to
(23:24):
take whatever they're doing dothe biggest possible version of
that.
When you combine that with,yeah, we're increasing the GP of
the internet.
All these products just kind ofstart to happen, I think.
So, so I think my last takeawayfor assembled has been to remind
people, like at some point withStripe, it was, it felt like, oh
(23:44):
yeah, like, what the heck, whatdoes increase the gp, the
internet?
We, but at some point then itbecame, yeah, we're gonna launch
more products.
This is not just, this is notthe end of it, this is not just
a second product.
It's not just a third product,not just a fourth product.
We're gonna do a ton ofproducts.
To do that.
And that's what we're trying todo with Assemble.
It's like, Hey, yeah, we'velaunched AI agents.
That's super cool.
It's super powerful.
It's a huge market.
(24:04):
And there's more coming afterthat.
Dan Balcauski (24:07):
So I heard it there.
Go hire the amortize and GregBrockman.
If you really
Ryan Wang (24:10):
right.
Dan Balcauski (24:11):
want bottom up transformation to successfully
work and a have a big vision.
Like, increasing the GDP of theinternet.
Well, hey, so, so I wanna goback to kind of the tactical.
So, the team is three, sixmonths in they've got some
traction now.
You're trying to take this thingto market.
I'm, walk me through like howyou thought about, okay, like
(24:31):
this is going to be a secondproduct.
We've got this workforcemanagement thing.
This is.
It's, you know, enablingcustomer support.
It fits under the broadumbrella.
But, I mean, those work well foremployees and maybe investors
company, customers don't alwayskind of understand how all
pieces fit together, nor do theyhave the patience for that in a,
30 minute demo pitch.
(24:52):
Right.
We're improving the world ofcustomer support.
So I guess like what happened,how are existing customers
reacting when you show up with anew a product AI product?
How did you think about,positioning that within, the
existing customers that youalready had.
Ryan Wang (25:06):
So the super interesting thing that was
unexpected to me was customersunderstood it immediately.
In the early days we'd hear alittle bit of, oh wait, I
thought you were workforcemanagement, but now you're doing
ai.
But then immediately the secondthing would be.
Oh, because you have all thedata, because you can route it,
whether it's a person or an AIor a BPO agent or somebody
(25:28):
you've hired, it's all costquality on the same curve, and
you're trying to deliver theend-to-end customer experience.
Like Yeah, that's what we'rejust gonna rip and put into our
deck what
Dan Balcauski (25:39):
Mm-hmm.
Ryan Wang (25:40):
exactly.
so customers understood it so,so, so early on and I think it
just goes back to the truism is.
When you talk to customers,magical things happen.
The harder part for us was tosay customers get it.
Some subset of the company getsit, but now we're kind of
(26:00):
turning on the go to marketengine once again.
And we have to, we can't skipsteps here because as a, later
stage company, you feel like,okay from the early days, you're
figuring out from the scalingdays, you're making it
repeatable.
Here's the playbook for how tolearn about call centers, how to
learn about workforcemanagement, what our product
does, who our competitors are,and we to go back to people and
say.
(26:20):
It's, there's no playbookanymore.
Yeah.
Take what you know, Gershwin andThia said and extrapolate that
into what David at DoorDashsaid, and put these things
together and mix and match.
And the set of people who arereally good at doing the
repeatability thing are notnecessarily the same people who
are really good at the mix andmatch thing.
Dan Balcauski (26:37):
It kind of almost reverts to like a founder led
sales again, right?
Where they talk about like earlydays of the product, like the
founder is gonna have to sellfor a while.
'cause there is not a repeatableway to pitch it until you sort
of learn and then you could sortof create the playbooks and hand
that off and, you know, buildout, you know, the whole sales
team.
Uh, so when you introduce like aproduct, you almost have that
problem again.
I mean, there's obviously amachine already running that
you'd like to just drop it into,but you don't have a pitch yet.
(27:00):
Quite as, you know, uh, likesteps 1, 2, 3, that sort of
cookie cutter, that maybe, youknow, your new brand new AE off
the street can just kind of slotinto.
Is that,
Ryan Wang (27:10):
Totally.
Dan Balcauski (27:10):
yeah.
Ryan Wang (27:11):
And then the founder-led sale is so important
on the second product.
But then and you just go all theway back to early days of, what
do you tell any company goingfrom zero to one, one to 10, et
cetera.
Well then you have to build outthe sales team.
Well, don't go hire a hundredpeople just because your model,
your spreadsheet model says youneed, this to, you gotta go one,
then two, then four, right?
(27:31):
Or, start with two, maybe thenfour, then eight, and then you
have, figure it out and build itup over time.
The.
The amnesia of success, so tospeak, is yeah, look, we've got
this sales team that they'rereally good at selling this
other thing.
Really understand the space,have all these relationships,
let's give all of them thisproduct.
No, you gotta start small againand build it back
Dan Balcauski (27:53):
Is that what you did?
Did you, I mean, did you try toroll it out to the whole sales
team?
What was actually the experienceof rolling out the ai agent
copilot product?
Did you, was that the step.
Ryan Wang (28:02):
tried to roll it out to the full sales team and it
didn't stick the first time.
We had some people who wereleaning in, some people who were
excited.
Some people who, still in thatkind of mindset of yeah, let me
figure it out.
This is amazing.
This is fun.
And some people who are, ooh,this is I need to figure, like,
I need some help here.
I need some data.
Meanwhile I'm doing a prettygood job selling the first
(28:23):
product.
So we ran into all of those.
And so we rolled it all the wayback and we got, a little bit
lucky in that there was one aewho.
Just was having the worst year.
Just I think he was all the waythrough the year, like September
or something, or October.
And had put n like, no, zeroclose.
(28:45):
It
Dan Balcauski (28:46):
Ooh.
Ryan Wang (28:47):
And so we said, Brian do you wanna sell the AI
product?
And just, you'll be thededicated person.
You'll figure out all thisstuff.
And I don't know if it was Exci,he's been very successful.
So I, we joke around with him,but it was like, I don't know if
he had a choice, it wasn't goingsuper well.
But he took to it.
Filled in a lot of theplaybooks.
He worked really closely withJohn, our co-founder.
(29:09):
And so they had this reallyamazing partnership of you could
sell a little bit ahead and themore they worked together, the
better it got.
And then he was able to bringthat back into the rest of the
team and they go 1, 2, 4, 8 andget it to everybody else because
you could see what successlooked like.
You had the playbooks and youhad this person who just felt
really confident once againselling the product'cause.
(29:31):
He was the only one.
So it was kind of like, shootingfish in a barrel.
Dan Balcauski (29:34):
I love that.
So you found maybe an unlikelycandidate, but somebody who
could lead the charge, pioneerthe process iterate, show
success, and then return thatback to the team to kinda show
what the.
What that pattern looked like,uh, versus trying to send a
hundred people all at once intothe jungle and have them all be
confused and half the teamdoesn't know what's going, which
(29:54):
way is up.
Love that part.
So, you know, I because I'm a ahuge nerd about this stuff, I
was looking at your site, and soyou do have, uh, a public
pricing page, which is awesome.
Um, and then you have your, uh,workforce management
capabilities separate from theai, uh, agent and copilot.
I'm curious okay, obviously, youknow, we've talked about them in
(30:16):
the sense of two products, butdidn't have to be that way.
Like how did you think aboutOkay.
We're seeing a bunch ofdifferent patterns in the
marketplace.
You know, folks adding, youknow, Microsoft with Office 365
has, their kind of co-pilot asan add-on to the main Office 365
subscription.
Uh, Gemini has bundled, all oftheir AI capabilities directly
into Google Workspacecapabilities.
(30:38):
Uh, I think it's pretty smart ontheir front.
And also they don't want tocompete head to head against,
like, if I spending$20 on ChachaPT or$20 on Gemini, I'll just
like, we'll just put it into thehigher tier plans of Google
Workspace.
How did you end up thinkingthrough the structure that
exists today?
Like what led you to thatdecision for your market?
Ryan Wang (30:56):
Yeah, it really did go back to customers once again.
So how do they buy in this case?
And what we found was weinternally and to investors, and
even to customers, wanted totell this amazing platform
story.
Hey.
It's human plus ai.
You've got customers coming inand route them to the right ai,
route them to the right person.
(31:17):
Or if they talk to the AI andthey wanna talk to a person, get
'em to a person, or you designit, but we give you all the
tools to design it.
we found that that wasn't thelead.
was the answer to the question.
Oh, there's a lot of AI agentsout there.
What's the difference betweenassembled end?
hundred other tools that you'vetalked to.
But it wasn't the lead.
And the lead was actually,there's a customer who they're
(31:41):
trying to put voice, AI voice infront of their one 800 number.
And otherwise, it's off.
It's just, it's only certainhours when you have people or
it's really expensive to serviceor you don't have it in certain
languages or so, so they werelooking for AI voice, or they
were looking for an AI chatbecause they're last gen set of
(32:02):
tools.
They're not gen ai, they arekind of linear.
Like you click the bubble, doyou want a refund or do you
wanna talk to, click the button.
So they wanted to replace that.
Or they were still looking forworkforce management.
Hey, we've got hundreds ofpeople.
We're not ready to do ai.
Or maybe we already have done AIreally quickly, but there's
still all these people, and nowwe're in a world of how do you
(32:23):
figure out the resourceallocation between people and
ai?
So they want workforcemanagement.
So it went back to how do we setup those multiple front doors
for each of these use cases,because that's what people are
trying to buy.
Then when they ask the question,what's the difference is when we
talk about the platform, thereis a different asterisk here,
which is when we're going toexisting customers.
(32:43):
When we're going to existingcustomers, then it's like, yeah,
hey, you already use assembledfor workforce management for ai,
copilot for AI chat.
We make it easy for them to.
To move the dollars around,frankly hey, we're thinking
about kind of total a CV totalTCP, contract value.
And I've been fascinated byMicrosoft's ability to have the
(33:05):
enterprise agreement, all youcan eat, whatever it is that you
want, you
Dan Balcauski (33:08):
To the chagrin of every company in the world,
including Slack, whicheventually had to find refuge in
Salesforce.
Ryan Wang (33:14):
exactly.
I think that's the goldstandard, right?
Just.
You tell us problems, you payus, you know enough for us to
solve'em and we'll just go solve'em, whatever they are.
'cause we do everything underthe sun.
Dan Balcauski (33:26):
I loved what you laid out there.
Just reflect back what I heardbecause what you said there was
we want to meet the market whereit is and the questions that
those prospects are coming to usto solve, not on necessarily how
we want to dictate to themarket, like our technology
platform upfront.
Because I see a lot of companiesgo.
(33:46):
The opposite way.
And it usually just ends up in aton of friction and a ton of
sadness.
And then the other thing yousaid, which I really like, I
agree, the new customer who'slooking at your pricing page,
very different from the existingcustomer.
Uh, one because that existingcustomer has a lot more
investment in hearing the fullstory.
(34:08):
Then the person who's visitingyour page, maybe they spend 30
seconds on the pricing page andmaybe they, you know, read a
couple of product pages.
But, you know, they don't havetime or the ability to sort of
comprehend the full thing.
They're just trying to get theirparticular questions answered.
And so it's a different model.
And so I think that's totally,good.
That's a perfectly validapproach so that you can have a
different pitch for the existingcustomer because they are
(34:30):
already involved in theecosystem and can kind of grasp
that more rich, platform story.
I look, there's a ton of thingswe didn't get to that I would
love to ask, but we are runningoutta time, so I wanna be
respectful of yours and theaudience.
I wanna wrap it out with acouple of rapid fire close out
questions.
You up for it.
Ryan Wang (34:46):
That's it.
Dan Balcauski (34:47):
Awesome.
Well.
When you think about all thespectacular people you've had a
chance to work with, is thereanyone who just pops to mind and
has had a disproportionateeffect on the way that you think
about building, runningcompanies?
Ryan Wang (34:59):
Yeah.
The person that comes to mind Ihas nothing to do with
technology, so a woman namedEmily Oster.
She now is well known forwriting a bunch of data-driven
books about parenting when I wasin high school and she was a
second year, I think, assistantprofessor at the University of
Chicago Department of Economics.
(35:19):
So.
Pretty busy time for a prettycompetitive place to be.
I emailed her and as well as abunch of other people at the
University of Chicago asking,Hey could you be a mentor on
this research project?
My, my school had this kookything where you could do
Wednesdays and you would go do aresearch project with somebody.
(35:41):
it, naturally most of theeconomists at, the Virgin
University of Chicago departmenteconomics were pretty busy and
did not respond.
And Emily did and she said, surething.
And come by my office and shemyself and my project partner
just sit outside her office andwe must have been 17 or 18 and
it, to this day, I still reflecton, wow, this person so busy.
(36:06):
So high powered, so much otherstuff going on, but took the
time outta the day to not justrespond to the email, but to
mentor us.
And so, there's that truism.
You send a good email and youfind that people are are pretty
generous with their time.
That was my experience and I tryto pay that forward a lot.
Dan Balcauski (36:20):
Well, props to Emily.
Yeah, it's amazing the impactyou can have on others when you
stop and give some of your time.
And there's also that truismthat like the busiest people are
the.
Easiest to reach.
It's like the,
Ryan Wang (36:31):
Right.
Dan Balcauski (36:32):
they there's either someone will answer email
in five minutes or they'll neverrespond.
And like the folks who answeredin five minutes also tend to be
the people who are like, I neverthought you would respond to
that email.
So props, props to Emily.
It made me think of, there's agreat book.
I've been in the consultingworld, a great book how to Talk
So Kids Will Listen.
It's a parenting book as well.
But it's awesome for aconsultant who's trying to speak
(36:54):
to clients because a lot of thesame principles apply.
And that's not to demean that'snot to mean clients in any way.
But oftentimes we don't make ourmessage heard.
In a way that can be heard bythe other party.
So, always a important lesson.
Uh, how do you stay sharp as aCEO?
What are you reading, listeningto learning from right now?
What any, uh, key sources thathave really changed your world
(37:15):
for you recently?
Ryan Wang (37:18):
Yeah I think two very different directions.
I very long ago.
When I was at the University ofChicago actually I took this
writing class and the writingclass was all about science
writing.
And because we were trying towrite like an honors paper or
something.
And the instructor, it was, Ithink it was called The Little
Red Schoolhouse.
(37:38):
It's a well-known class in abook.
And he was explaining actuallykind of similar to you, how do
you write for the audience?
And a lot of the times you writethis way because that's how you
think about it when you really,you have to explain it this way
'cause that's how the audienceis thinking about it.
anyway, he said the Nobel Prizeis an amazing example of this.
You go pull up the prizeannouncements for.
For physics, for biology, forchemistry, et cetera, et cetera.
(38:02):
And you find that there's threeversions of every announcement.
There's the public pressrelease, there's the technical
background, and then there's thespeech.
Same topic, very differentlevels.
Dan Balcauski (38:12):
Hmm.
Ryan Wang (38:13):
so I find that reading through these both as
interesting in an exercise ofhow do you dumb down concepts,
not dumb down, explain indifferent ways.
And just keeps, it's justinteresting.
It keeps me thinking about likewhat is happening in the world
that's coming not just nextyear, two years from now, but 10
years from now, 15 years fromnow, 20 years from now.
The total other direction fromthat is I love the 20 VC
(38:33):
podcast.
They have this group panel nowwith R Driscoll and Harry
Stabbings and Jason Lemkin, andthey just talk a lot about
current affairs and specificallywith this very deep lens on
technology.
It's very kind of turpentinetactical, which is what's useful
for me in the very, very nearterm.
Dan Balcauski (38:50):
Nice.
Nice.
Well, if I give you a billboard,you can put any advice on there.
For other B2B SaaS CEOs tryingto scale their companies, what
would it say?
Ryan Wang (38:58):
I.
I think I'd say, why are youpaying for billboards?
Get on the plane.
But seriously, I do think mybelief is that, you win
customers fundamentally by wordof mouth.
Yeah.
Awareness helps goose that forsure.
But at the end of the day,there's nothing that beats
finding somebody, having themtake a bet on you.
Doing a really good job for themand them telling everybody else
(39:20):
that they know, like, Hey, wow.
Assembled their AI agent ortheir workforce management
platform.
Here's what they did, and thenpercolate that out.
And then I found that all of thegreat companies started there is
do a really good job.
And then your best marketing isyour best awareness is the
people who you've already workedwith.
Dan Balcauski (39:39):
Do a great job and get on the plane.
Love it.
This has been fantastic, Ryan.
If our listeners want to connectwith you, learn more about
assembled, how can they do that?
Ryan Wang (39:46):
We're at Assembled.com.
We post a lot of stuff on ourLinkedIn, especially around the
intersection of AI and customersupport.
It's best place to check us out.
Dan Balcauski (39:54):
Awesome.
Well, I'll put those links inthe show notes for listeners.
Everyone that wraps up thisepisode of Sask Scaling Secret.
Thank you to Ryan for sharinghis journey and insights.
For listeners, you found Ryan'sinsights valuable.
Please leave a review and sharethis episode with your network.
Really helps the podcast grow.
Yeah it started as a second product and then we
(00:02):
accidentally turned into three,four, and five all at once.
in 2022, of course chat GTlaunched and.
With our ML backgrounds and justkind of surveying the technology
landscape it became reallyobvious to us, well, what is
possible in customer support hastotally changed and we should
not build AI for support becauseit's cool or because the
(00:24):
technology seems really fun toplay around with.
But it actually did start intruth with somebody at a
hackathon putting together a botand dropping into Slack.
When you talk to customers,magical things happen.
my belief is that, you wincustomers fundamentally by word
of mouth.
Awareness helps but at the endof the day, there's nothing that
beats finding somebody, havingthem take a bet on you.
Dan Balcauski (01:08):
Welcome to SaaS Scaling Secrets, the podcast
that brings you the insidestories from the leaders of the
best scale up.
B2B SaaS companies.
I'm your host, Dan Balcauski,founder of Product Tranquility.
Today I'm excited to welcomeRyan Wang, CEO and co-founder of
assembled a support operationsplatform that helps companies
like Stripe, zoom, and Etsymanage their customer support at
scale.
Before founding assembled, Ryanwas employee number 80 at
(01:28):
Stripe, where he watched thecompany scale from 80 to 800
people and helped build the MLsystems that powered fraud
detection and supportautomation.
Ryan, welcome to the show.
Ryan Wang (01:37):
Great to be.
Dan Balcauski (01:39):
Very excited for our conversation today.
Ryan, before we dive into yourscaling journey, just give us
the elevator pitch.
What does Assembled do?
Who do you serve?
Ryan Wang (01:46):
Yeah, assembled is the AI platform for customer
support.
So we help companies with thefull gamut of all things
customer service.
So automation to solve 70, 80%of your customer tickets over
chat, over voice, over emailaugmentation.
So co-pilot that makes people20, 30% more
Dan Balcauski (02:02):
I.
Ryan Wang (02:02):
ramp 50% faster.
All the way through tooperations.
So helping companies manage20,000 plus person contact
centers with the forecasting andthe staffing planning and the
capacity planning betweenultimately whether you should
talk to an AI agent or whetheryou should talk to a human agent
or whether that should beinterspersed, making sure your
customers have the rightexperience.
We help companies of alldifferent shapes and sizes, so
(02:23):
Ashley Furniture, Stripe, RobinHood.
So, so it's quite the complexset of customer service
concerns.
Dan Balcauski (02:32):
Well quite a broad portfolio and I wanna dive
into that set of capabilitiesdirectly.
So as I understand it, youcorrect me if I'm wrong, but
assembled started as a workforcemanagement.
Platform and now you've addedthese AI agent and copilot
capabilities.
And so adding that, maybe youconsider it a second product or
(02:52):
not, but it kind of appears thatway, could often be a
challenging transition forscaling companies.
So I'm really curious about thedecision around adding.
Additional product becauseagain, this could be a fraught
time.
Um, what was your thoughtprocess in terms of timing?
Like how did you think aboutthis is the right time to do it
(03:13):
or not?
And we go from there.
Ryan Wang (03:15):
Yeah it started as a second product and then we
accidentally turned into three,four, and five all at once.
And so really stretched ourroadmap.
The story goes further back.
So I was at Stripe.
I was actually working onmachine learning for fraud
detection.
This was around 2014 or so.
And had nothing to do withcustomer service, but Patrick
and John Collison, theco-founder of Stripe.
(03:37):
They cared about it a lot.
They would do support rotations.
They would have people over totheir apartment in the mission
in San Francisco and invite thewhole company to do customer
support.
And then over time, of course,from 80 to 800, that doesn't
make sense.
It doesn't scale.
You bring in outsourcers, youstart to have different products
that you're servicing acrossdifferent geographies.
(03:59):
You start to need to havedifferent tiers of escalation.
And so, it became really obviousright around 20 17, 20 18, that
customer support was one ofthose spaces where you would
apply, not ai.
But machine learning and it wasin service of helping the
company scale, great customersupport, being high quality,
(04:19):
really efficient multi-product,global support.
And so when we started, ourfirst product was workforce
management.
But that actually came out of aquestion that we asked to the
folks at Stripe.
There was a guy named Bob VanWinden who ran support and he
had come from Google and seenthe challenges scaling up
Google, and we asked him justWhat is software you need to buy
(04:42):
or you hit your options.
And it wasn't just Bob, it was ahundred other support leaders
but a lot of them came back withthis answer of, well.
This thing called workforcemanagement.
It's a team that's growing from,several hundred to several
thousand and figuring out whoshould be working on what,
getting the right person in theright place at the right time.
That was the challenge of 2018in terms of scaling support and
(05:02):
exacerbated in 2020 when welaunched.
So it actually took us two yearsto get off the blocks.
But in 2020, companies weregrowing really quickly, and it
was all through headcount.
And so workforce managementbecame an even bigger problem.
And then in 2022, of course chatGT launched and.
With our ML backgrounds and justkind of surveying the technology
landscape it became reallyobvious to us, well, what is
(05:23):
possible in customer support hastotally changed and we should
not build AI for support becauseit's cool or because the
technology seems really fun toplay around with.
But because that answer to thatquestion, how do you scale great
customer support.
We feel like has changed in thismoment.
So, it was really easy for us tosay, yeah, that's something we
(05:44):
wanna do.
It's something that fits ourmission.
It's something that fits ourproduct portfolio in terms of
the value we deliver and andthen stack these products
together.
Dan Balcauski (05:53):
So, so solving the same core problem and this
happenstance of the technologymeeting capabilities with your
background experience, uh,around, you know, the ability to
bring ML to really solve thisproblem well outside of just
adding humans to the mix.
So you guys decide to go fullbore on this.
(06:15):
How did you approach it?
Like meaning, how did you thinkabout organizing the team to
build this additional set ofcapabilities?
And maybe at that time it wasmaybe, I dunno if you thought of
it originally as a net newproduct or just a feature set,
but kinda lead me through thatprocess.
Ryan Wang (06:33):
Yeah we did think about it as a new product from
the very beginning, I will sayit was.
2.5, possibly take three ofbuilding a new product that
assembled and so company hadactually come out of this
bottoms up motion at Stripe.
We were building ML tools forcustomer support.
But I can tell you.
(06:54):
asked us to do it.
They actually, they asked us notto do it.
I remember having this meetingwith the head of product and the
head of systems engineering andthe head of product said, Hey,
you should be working on thedashboard.
That's a high priority.
There's not enough peopleworking on that.
Why are you working on support?
And the head of systemsengineering said, you're doing
machine learning for customersupport.
Doesn't NLP solve all this,kinda shades of AI doesn't like
(07:15):
what's the point?
Isn't this just kind of it onand it works?
So, so actually the firstproject within Stripe that was
kind of ML support happened.
It was almost a startup.
Within a startup.
It was.
myself and one of my co-foundersBrian Z trying to solve a
problem, going to visit callcenters, figuring out what their
problems were trying to pull inmore engineers and scrap for
(07:39):
resources.
We borrowed this intern, said,Hey, we have this project going
on and it's starting to delivera lot of impact and we think it
could be a really big teameventually.
So that was our mindset cominginto assembled of, okay, well we
had created a product and a lotand that wasn't just customer
support.
A lot of the great products thatStripe had come bottoms up
somebody.
There's this thing called theCrazy Ideas Hackpad.
(08:02):
looked on the crazy ideasHackpad, and they were hired to
do this, but they said, actuallyit would be cool to have an API
to create a company.
That was how Shred Ball cameabout.
So our first.
product.
We just thought, hey, if we tellpeople, you know what, here's
the story of how assembled evengot started and if you see the
right opportunity, you don'thave to ask, just do it.
(08:23):
Create a new product.
And it never off the got off theground.
There was a group of people thatwould meet weekly and they were
trying to build a product calledRoute Automated Routing, and now
it's part of our productportfolio.
But they would meet every weekand it just never caught.
And then our second attempt at aproduct.
It turned out to be a killerfeature.
It was a feature to help BPOs oroutsourcers connect into
(08:46):
customers like many of ourlargest customers.
So, so if you have multipleBPOs, it's almost like a network
of.
Rather than individuals,hundreds of individuals at a
time.
So we wanted to give peoplevisibility in that whole network
rather than just the people.
And we came super tops down.
We wrote a strategy.
A woman who is an MBA internwrote a really impressive
(09:09):
document on all the BPOs weshould go after in what order,
the integrations we would needthe size of the market.
She's now at OpenAI.
So it was really high quality,but just took us so long to get
off the blocks there.
And then the minute we startedbuilding it was like, hmm, the
strategy's not quite right.
Actually, we gotta start withthe integrations, not the BPOs
this way.
So our new product around AI waskind of the synthesis of those
(09:31):
two things was.
We know strategically why we'redoing this.
We want to help deliver greatcustomer support.
We know the technology LMS aregonna change everything in terms
of what's possible.
But it actually did start intruth with somebody at a
hackathon putting together a botand dropping into Slack.
And we had this Slack channelcalled product Questions.
(09:55):
People come ask, Hey, how doesour in Workday integration work?
And in truth, it was, product,70% of the time you get an
answer to your question.
but when this bot showed up andit was automatically answering
questions, our support team, oursales team, our customer success
team, they were, oh my gosh,we're actually getting answers
to our questions, and they'reright and they're good.
(10:16):
So from there it became reallyobvious, okay, we have
something.
There's something to buildaround.
There's a use case.
And then the top down version oftaking that idea and running
with it was to say, we're gonnaput you in a separate room this
time around.
We're gonna put you in adifferent Slack group where
you're gonna, gonna have peoplededicated to this team and your
(10:36):
whole mission, and your wholegoal is to create a new product.
And we're gonna borrow theplaybook.
Even of Y Combinator, you have12 weeks.
At the end of 12 weeks, you'regonna do a demo day.
And.
The fast forward is that itdidn't take 12 weeks.
It took a little bit longer than12 weeks, but it operated in
that kind of cycle.
Dan Balcauski (10:54):
So just to, so I could lay out the roadmap for
listeners a little bit.
So, the AI product not thesecond product necessarily, but
you described it, maybe it's,2.5 or three, depending upon how
you look at it with thisinterim, products.
That was for the BPO.
So you have sort of workforcemanagement, uh, this BPO, uh,
initiative, and then you havethe new AI agent and copilot
(11:17):
offering.
Now I'm curious you, so you saidthe, you, so you separate this
group off.
Y Combinator style to go run for12 ish weeks.
What were you trying to avoidwith that situation?
I guess what was the lessonlearned from that intermediate
time where you're like, oh, thiswill be the better way to go?
Ryan Wang (11:37):
Yeah.
Yeah.
The lesson from the firstexperiment was that it needed to
be a dedicated team with a cleargoal.
Dan Balcauski (11:43):
Hmm.
Ryan Wang (11:43):
it couldn't just be purely bottoms up.
And so, for example.
We, we didn't tap people.
We didn't say, Hey, you're, oneof our strongest engineers.
Go join this team.
Instead.
We had an application process.
The application was just threeshort questions.
It wasn't super formal.
Have you done at assembled sofar?
(12:04):
why do you wanna join this team?
And what do you know about ai?
you could kind of, tease apartmost of the characteristics that
we need in the team from thosequestions.
But the lesson from the secondangle of getting too hung up in
strategy, not going too topsdown was just, it's the classic
Mike Tyson quote.
Everybody's got a plan untilthey're punched in the face.
(12:25):
We had this plan, we just don'tknow.
The space is evolving, thetechnology's evolving, and then
also in the early days ofassembled, it's not like we set
out and said.
We're going to go solveworkforce management.
what we said is we're going togo improve customer support.
That's the broad bucket in whichwe play.
And then we're gonna go ask abunch of people who know a lot
about customer support, what isthe specific problem?
(12:45):
So we wanted to put people intothat situation where.
could just ask that question,run after it, learn really
quickly, be unencumbered by aplan that was to set based on
things that, that they didn't,we didn't learn from customers.
And then have a clear goal atthe end of the day.
Dan Balcauski (13:02):
So, okay, so you put these folks off Tiger Team
Isolation dedicated process witha high maybe some sort of filter
into that group that, you know,these folks are qualified
go-getters, uh, that you want inthere.
So you know, they go run off.
I guess, what was your idea for,okay, like they're gonna go
create something.
(13:22):
How do you, was the idea like,okay we'll then be just kind of
creating like a, another companydivision.
Was it like, what was thethought about like how you're
going to reintegrate them, Iguess?
Tell me what that looked likewhen, what's that 12 ish week
period is done?
Ryan Wang (13:38):
Yeah.
Well this part we really messedup, I would say.
And when you read the.
HVR McKinsey articles aboutHorizon 1, 2, 3.
They make it sound so clean.
You're in the zero to one phaseand then you know when you're
out and then you put it in thisother stage.
It just.
Maybe we didn't read thearticles closely enough, but we
definitely messed up on thereintegration part.
(13:59):
I think because it took a reallylong time for it to be clear
that it would be successful.
In truth, at the end of the 12week demo day there was some
really exciting progress.
There was some, adoption chartsthat were going up, but it
wasn't this is in 2023.
It wasn't like, oh geez.
E everything sorted out and youcould see this thing going from
zero to a hundred million a rovernight.
(14:20):
It was, you kind of had tosquint a little bit
Dan Balcauski (14:23):
Mm.
Ryan Wang (14:23):
There was early customers where, okay, they're
really sticking with what we'redoing, but we have to keep
iterating here so.
The 12 week demo day turnedinto, you know what, we'll check
back in again in three months.
And then instead of it justbeing a demo day for a small
group of people, you're gonna goin front of the whole company.
It'll be at the all hands.
And so, you'll have to do thatpitch for everybody.
(14:46):
And even then, I think.
The mistake was we had kind ofset it up as the company is
almost like judging you,American Idol style it created a
little bit of resentment.
come these people who, filledout the application, they got to
be on the team and we're overhere.
Working on the old thing.
They get to mess around with AIand it's super fun and they were
(15:09):
having a lot of fun in the room.
I was like, how come they'rehaving so much fun?
This is what the heck, we'reover here solving big,
multimillion dollar a yearcustomer problems, and they're
over there just throwing stuffat the wall and moving fast.
The hell, so, so that resentmentlingered I think when it became
really clear, okay, this issuccessful.
We want more to put moreinvestment for us of the
(15:29):
company.
We wanna move people from ourolder product to our newer
product.
And we get this question all thetime.
Are we pivoting the company?
Are we pivoting the company andhow do these products go
together and should we jet us inthis off?
Or the small group said, maybewe should turn this into a
separate startup and it's asubsidiary
Dan Balcauski (15:46):
Mm.
Ryan Wang (15:46):
And in both directions.
And that resentment was reallyproblematic because the story we
tell about assembled that I justtold you is.
Well, our value prop is to behuman and ai.
It's to own the end-to-endcustomer experience.
And all the products on top ofthat, like these products are
stronger together and they'reall about helping our customers
deliver better customer support,high quality, super efficient.
You can't do that if it's justone piece of this kind of narrow
(16:09):
slice.
But we really, really had tobring people back into that
because of, I think some of thisresentment, some of this
cultural built up story.
And then also because we wentinto it without a plan.
Dan Balcauski (16:21):
Mm-hmm.
Ryan Wang (16:21):
without a plan, but without, but by saying, who
cares about the strategy?
We'll figure it out after it'ssuccessful.
But then once it was successful,we really had to build the story
back.
Dan Balcauski (16:29):
I, well, I'm curious.
Uh, it's funny because I'veactually, uh, my previous life
as a product leader, I got putin one of these labs teams as
like, you're gonna horri,basically Yeah.
These Horizon three group.
And so I've lived that.
World of resentment.
The irony is, is like it's notany better on the other side.
Like, like, like they're alwayshaving fun over there.
(16:52):
They caught you in a moment oflike gallows humor probably
because like you're over therebeating your head against the
wall, being like, it's still notworking.
It's still not working.
Ryan Wang (17:00):
Yeah.
Dan Balcauski (17:00):
Our time is running out budget is running
out.
Like we've got show progress.
I'm curious, like as a leaderyou know what, I guess you know,
was that.
People, people's feelings Iguess are always valid.
Do people have feelings?
But I'm sure like there was aperception issue.
Like are there things that, youfound that were successful in
diffusing that, or things thatguys looking back that you like
(17:22):
would have like implemented?
Like just think about otherpeople listening if they're
gonna like, send off these cyberteams.
Right.
Like it's interesting'cause youalways think about like, okay,
what's the market acceptance ofthis thing?
And I could easily see how thatsort of cultural resentment.
Maybe like it was like, oh, itwasn't even on the checklist of
things I needed to worry about,and all of a sudden I'm having
to manage that.
So like as you think about that,reflect on that time as a leader
now, like is there things thatyou did to help push that
(17:45):
forward or, even if it was laterthan you might have anticipated.
Ryan Wang (17:49):
Well, I resonate a lot with what you said.
It feels like outside lookingin, it's so fun.
And then inside looking out andthis was the resentment going
the other way that we saw too.
Like, what the heck?
This is really hard.
It's existential, it's like wedon't even have any customers
yet.
Or, Hey, we have like threecustomers.
We're trying to get to 20.
How come the rest of the companywon't help us?
And I think a lot of the re theliterature, there's so much
(18:09):
great literature now on how togo multi-product, the mechanics
of it, and when to do it and whyto do it earlier.
There's not much on thiscultural piece for me.
I think lesson one or the first,most useful thing, and I, it
wasn't necessarily that I wentin prepared.
I talked to this guy, PeterGasner, who started a company
called Veeva,$50 billion LifeScience SA Company.
(18:33):
And this was about a, this wasin 2022.
Or so so before we even startedthis initiative, and he had told
me in the early days of Viva,they built a second product and
intentionally he said, we pullit as far from our first one as
humanly possible.
Why?
Because the tendency is to govery iterative.
(18:53):
So in Viva, in the early days,second product as far away as
possible, from the first one hesaid, we almost broke the
company.
This is, we almost broke thecompany and he elude.
I didn't understand at the time,but he alluded to the reasons it
was, I think because of thecultural pieces.
It's these questions.
Yeah.
Are we pivoting?
We thought we were workforcemanagement.
No, we, this company like.
(19:14):
joined the company and thatproduct was working.
No, you joined the company.
I thought we were improvingcustomer support.
Full stop.
So you have to lift people's eyeline.
You have to show them, Hey,yeah.
Dan Balcauski (19:24):
Mm-hmm.
Ryan Wang (19:24):
today's world, in any given moment, there's
constraints.
is competing against that interms of resources.
Go here or there.
But in the long term, all thesethings are coming back together
to be a powerful platform.
And you really have to showpeople that way.
And I do think that's partstorytelling on the part of the
leadership.
That was, I would say the secondpart.
And that contributes to thecultural story of why are we
(19:46):
doing this?
Are we doing this because weneed more a RR?
Are we doing this because AI issuper hot?
Are we doing this?
Because the other thing's notworking.
No, we're doing this becausethese things are better
together.
But that's a big part of keepingpeople motivated and them
understanding the story andseeing it for themselves.
Even though, again, just becauseof the, how far apart these
things are, they might have tosquint a little bit and they
(20:08):
might not get it for a while.
Dan Balcauski (20:10):
Well, I mean, uh, is the old, uh, adage, nature of
bores a vacuum and humans, ifthere's no information, we tend
to paint in that black spot ourworst visions of it.
Right.
So what you mentioned is whatwe're doing not working?
Or is this like another bet thecompany thing to get us, you
know, you're like, no, no, no.
It's all, all in the same plan,but we've gotta experiment.
(20:30):
So I could see how that be achallenge.
You mentioned before, I wanna.
We'll come back to kind of,launching with this new product,
but I want to just tie it backbecause you did talk a little
bit about Stripe and how StripeAtlas came out of this, kind of,
sort of bottoms up and I guesswhat.
What do you feel like was,either different in the water of
what you saw at Stripe versuswhat you experienced at
(20:52):
assembled?
I mean, maybe you were just in adifferent position as a, you
know, if you're, you know, anengineer kind of working on
stuff versus now you're the CEleader.
Maybe, you know, maybe the Coonshad the exact same view of like,
dysfunction, right?
Of like, oh my God, there's abunch of people launching stuff
and nothing's connected.
I don't know, I haven't pickedtheir brain.
Um, be happy to talk to'em atsome point, but, um, I'm
curious, like, kind of lookingback like what do you, like, was
(21:13):
it a cultural thing?
Was it like part of the systemsand processes they had for kind
of taking those products fromincubation into the market?
Like what do you what was yoursense of like maybe what was
different?
Ryan Wang (21:24):
Yeah, I think the process and systems came later,
but I think two big dimensionsthat worked in concert, I think
one was that kind of clarity ofmission.
That was a really big one.
And Stripe, they're not apayments company.
It's increasing the GDP of theinternet.
(21:46):
And when I was an engineer, Ithought, that's so stupid.
What, what does that mean?
Increase the GDP of theinternet?
No.
What?
Like, let's build technology,let's improve payments.
I don't get it.
but now when you zoom though allthe way out, it's like, how does
Bridge Stablecoin fit in?
How does Stripe Atlas fit in?
How does having.
(22:06):
printing press, you know that.
Does books make sense?
Well, yeah.
It's not a payments company.
It's increasing the GP of theinternet.
All of these bets fit under thatbig tent.
And similar for us, I thinkwe're trying to improve customer
support, full stop.
But even beyond that, we wannasolve the operational challenges
that come after big ideas.
For companies like Stripe thatbecome successful.
Support is a function.
(22:27):
Trust and safety is a function.
Operations broadly is stuff thatcomes after.
So we try to fit people intothis.
It's not a workforce managementcompany, that's one of our
products for sure, but we'retrying to solve this very big
problem.
Dan Balcauski (22:39):
Hmm.
Ryan Wang (22:39):
I think the other part that fits that plays well
with a big vision is.
When, what the best people inthe world.
They don't wanna just, yeah, Idon't want, just wanna make a,
10% better payments company thanthe thing that came before.
Don't wanna just build customersupport systems and solve this,
legacy problem that's a littlebit better.
They wanna do big things.
And just to put that inperspective, my recruiter at
(23:02):
Stripe, Daniella Amedee, she'spresident of Anthropic my.
Onboarding was done by the thenCTO of Stripe.
Greg Brockman, he's of OpenAI,and so you know, you get this
density of people together whoare not just super, super smart,
super, super humble, all ofthose things, but just want to
(23:24):
take whatever they're doing dothe biggest possible version of
that.
When you combine that with,yeah, we're increasing the GP of
the internet.
All these products just kind ofstart to happen, I think.
So, so I think my last takeawayfor assembled has been to remind
people, like at some point withStripe, it was, it felt like, oh
(23:44):
yeah, like, what the heck, whatdoes increase the gp, the
internet?
We, but at some point then itbecame, yeah, we're gonna launch
more products.
This is not just, this is notthe end of it, this is not just
a second product.
It's not just a third product,not just a fourth product.
We're gonna do a ton ofproducts.
To do that.
And that's what we're trying todo with Assemble.
It's like, Hey, yeah, we'velaunched AI agents.
That's super cool.
It's super powerful.
It's a huge market.
(24:04):
And there's more coming afterthat.
Dan Balcauski (24:07):
So I heard it there.
Go hire the amortize and GregBrockman.
If you really
Ryan Wang (24:10):
right.
Dan Balcauski (24:11):
want bottom up transformation to successfully
work and a have a big vision.
Like, increasing the GDP of theinternet.
Well, hey, so, so I wanna goback to kind of the tactical.
So, the team is three, sixmonths in they've got some
traction now.
You're trying to take this thingto market.
I'm, walk me through like howyou thought about, okay, like
(24:31):
this is going to be a secondproduct.
We've got this workforcemanagement thing.
This is.
It's, you know, enablingcustomer support.
It fits under the broadumbrella.
But, I mean, those work well foremployees and maybe investors
company, customers don't alwayskind of understand how all
pieces fit together, nor do theyhave the patience for that in a,
30 minute demo pitch.
(24:52):
Right.
We're improving the world ofcustomer support.
So I guess like what happened,how are existing customers
reacting when you show up with anew a product AI product?
How did you think about,positioning that within, the
existing customers that youalready had.
Ryan Wang (25:06):
So the super interesting thing that was
unexpected to me was customersunderstood it immediately.
In the early days we'd hear alittle bit of, oh wait, I
thought you were workforcemanagement, but now you're doing
ai.
But then immediately the secondthing would be.
Oh, because you have all thedata, because you can route it,
whether it's a person or an AIor a BPO agent or somebody
(25:28):
you've hired, it's all costquality on the same curve, and
you're trying to deliver theend-to-end customer experience.
Like Yeah, that's what we'rejust gonna rip and put into our
deck what
Dan Balcauski (25:39):
Mm-hmm.
Ryan Wang (25:40):
exactly.
so customers understood it so,so, so early on and I think it
just goes back to the truism is.
When you talk to customers,magical things happen.
The harder part for us was tosay customers get it.
Some subset of the company getsit, but now we're kind of
(26:00):
turning on the go to marketengine once again.
And we have to, we can't skipsteps here because as a, later
stage company, you feel like,okay from the early days, you're
figuring out from the scalingdays, you're making it
repeatable.
Here's the playbook for how tolearn about call centers, how to
learn about workforcemanagement, what our product
does, who our competitors are,and we to go back to people and
say.
(26:20):
It's, there's no playbookanymore.
Yeah.
Take what you know, Gershwin andThia said and extrapolate that
into what David at DoorDashsaid, and put these things
together and mix and match.
And the set of people who arereally good at doing the
repeatability thing are notnecessarily the same people who
are really good at the mix andmatch thing.
Dan Balcauski (26:37):
It kind of almost reverts to like a founder led
sales again, right?
Where they talk about like earlydays of the product, like the
founder is gonna have to sellfor a while.
'cause there is not a repeatableway to pitch it until you sort
of learn and then you could sortof create the playbooks and hand
that off and, you know, buildout, you know, the whole sales
team.
Uh, so when you introduce like aproduct, you almost have that
problem again.
I mean, there's obviously amachine already running that
you'd like to just drop it into,but you don't have a pitch yet.
(27:00):
Quite as, you know, uh, likesteps 1, 2, 3, that sort of
cookie cutter, that maybe, youknow, your new brand new AE off
the street can just kind of slotinto.
Is that,
Ryan Wang (27:10):
Totally.
Dan Balcauski (27:10):
yeah.
Ryan Wang (27:11):
And then the founder-led sale is so important
on the second product.
But then and you just go all theway back to early days of, what
do you tell any company goingfrom zero to one, one to 10, et
cetera.
Well then you have to build outthe sales team.
Well, don't go hire a hundredpeople just because your model,
your spreadsheet model says youneed, this to, you gotta go one,
then two, then four, right?
(27:31):
Or, start with two, maybe thenfour, then eight, and then you
have, figure it out and build itup over time.
The.
The amnesia of success, so tospeak, is yeah, look, we've got
this sales team that they'rereally good at selling this
other thing.
Really understand the space,have all these relationships,
let's give all of them thisproduct.
No, you gotta start small againand build it back
Dan Balcauski (27:53):
Is that what you did?
Did you, I mean, did you try toroll it out to the whole sales
team?
What was actually the experienceof rolling out the ai agent
copilot product?
Did you, was that the step.
Ryan Wang (28:02):
tried to roll it out to the full sales team and it
didn't stick the first time.
We had some people who wereleaning in, some people who were
excited.
Some people who, still in thatkind of mindset of yeah, let me
figure it out.
This is amazing.
This is fun.
And some people who are, ooh,this is I need to figure, like,
I need some help here.
I need some data.
Meanwhile I'm doing a prettygood job selling the first
(28:23):
product.
So we ran into all of those.
And so we rolled it all the wayback and we got, a little bit
lucky in that there was one aewho.
Just was having the worst year.
Just I think he was all the waythrough the year, like September
or something, or October.
And had put n like, no, zeroclose.
(28:45):
It
Dan Balcauski (28:46):
Ooh.
Ryan Wang (28:47):
And so we said, Brian do you wanna sell the AI
product?
And just, you'll be thededicated person.
You'll figure out all thisstuff.
And I don't know if it was Exci,he's been very successful.
So I, we joke around with him,but it was like, I don't know if
he had a choice, it wasn't goingsuper well.
But he took to it.
Filled in a lot of theplaybooks.
He worked really closely withJohn, our co-founder.
(29:09):
And so they had this reallyamazing partnership of you could
sell a little bit ahead and themore they worked together, the
better it got.
And then he was able to bringthat back into the rest of the
team and they go 1, 2, 4, 8 andget it to everybody else because
you could see what successlooked like.
You had the playbooks and youhad this person who just felt
really confident once againselling the product'cause.
(29:31):
He was the only one.
So it was kind of like, shootingfish in a barrel.
Dan Balcauski (29:34):
I love that.
So you found maybe an unlikelycandidate, but somebody who
could lead the charge, pioneerthe process iterate, show
success, and then return thatback to the team to kinda show
what the.
What that pattern looked like,uh, versus trying to send a
hundred people all at once intothe jungle and have them all be
confused and half the teamdoesn't know what's going, which
(29:54):
way is up.
Love that part.
So, you know, I because I'm a ahuge nerd about this stuff, I
was looking at your site, and soyou do have, uh, a public
pricing page, which is awesome.
Um, and then you have your, uh,workforce management
capabilities separate from theai, uh, agent and copilot.
I'm curious okay, obviously, youknow, we've talked about them in
(30:16):
the sense of two products, butdidn't have to be that way.
Like how did you think aboutOkay.
We're seeing a bunch ofdifferent patterns in the
marketplace.
You know, folks adding, youknow, Microsoft with Office 365
has, their kind of co-pilot asan add-on to the main Office 365
subscription.
Uh, Gemini has bundled, all oftheir AI capabilities directly
into Google Workspacecapabilities.
(30:38):
Uh, I think it's pretty smart ontheir front.
And also they don't want tocompete head to head against,
like, if I spending$20 on ChachaPT or$20 on Gemini, I'll just
like, we'll just put it into thehigher tier plans of Google
Workspace.
How did you end up thinkingthrough the structure that
exists today?
Like what led you to thatdecision for your market?
Ryan Wang (30:56):
Yeah, it really did go back to customers once again.
So how do they buy in this case?
And what we found was weinternally and to investors, and
even to customers, wanted totell this amazing platform
story.
Hey.
It's human plus ai.
You've got customers coming inand route them to the right ai,
route them to the right person.
(31:17):
Or if they talk to the AI andthey wanna talk to a person, get
'em to a person, or you designit, but we give you all the
tools to design it.
we found that that wasn't thelead.
was the answer to the question.
Oh, there's a lot of AI agentsout there.
What's the difference betweenassembled end?
hundred other tools that you'vetalked to.
But it wasn't the lead.
And the lead was actually,there's a customer who they're
(31:41):
trying to put voice, AI voice infront of their one 800 number.
And otherwise, it's off.
It's just, it's only certainhours when you have people or
it's really expensive to serviceor you don't have it in certain
languages or so, so they werelooking for AI voice, or they
were looking for an AI chatbecause they're last gen set of
(32:02):
tools.
They're not gen ai, they arekind of linear.
Like you click the bubble, doyou want a refund or do you
wanna talk to, click the button.
So they wanted to replace that.
Or they were still looking forworkforce management.
Hey, we've got hundreds ofpeople.
We're not ready to do ai.
Or maybe we already have done AIreally quickly, but there's
still all these people, and nowwe're in a world of how do you
(32:23):
figure out the resourceallocation between people and
ai?
So they want workforcemanagement.
So it went back to how do we setup those multiple front doors
for each of these use cases,because that's what people are
trying to buy.
Then when they ask the question,what's the difference is when we
talk about the platform, thereis a different asterisk here,
which is when we're going toexisting customers.
(32:43):
When we're going to existingcustomers, then it's like, yeah,
hey, you already use assembledfor workforce management for ai,
copilot for AI chat.
We make it easy for them to.
To move the dollars around,frankly hey, we're thinking
about kind of total a CV totalTCP, contract value.
And I've been fascinated byMicrosoft's ability to have the
(33:05):
enterprise agreement, all youcan eat, whatever it is that you
want, you
Dan Balcauski (33:08):
To the chagrin of every company in the world,
including Slack, whicheventually had to find refuge in
Salesforce.
Ryan Wang (33:14):
exactly.
I think that's the goldstandard, right?
Just.
You tell us problems, you payus, you know enough for us to
solve'em and we'll just go solve'em, whatever they are.
'cause we do everything underthe sun.
Dan Balcauski (33:26):
I loved what you laid out there.
Just reflect back what I heardbecause what you said there was
we want to meet the market whereit is and the questions that
those prospects are coming to usto solve, not on necessarily how
we want to dictate to themarket, like our technology
platform upfront.
Because I see a lot of companiesgo.
(33:46):
The opposite way.
And it usually just ends up in aton of friction and a ton of
sadness.
And then the other thing yousaid, which I really like, I
agree, the new customer who'slooking at your pricing page,
very different from the existingcustomer.
Uh, one because that existingcustomer has a lot more
investment in hearing the fullstory.
(34:08):
Then the person who's visitingyour page, maybe they spend 30
seconds on the pricing page andmaybe they, you know, read a
couple of product pages.
But, you know, they don't havetime or the ability to sort of
comprehend the full thing.
They're just trying to get theirparticular questions answered.
And so it's a different model.
And so I think that's totally,good.
That's a perfectly validapproach so that you can have a
different pitch for the existingcustomer because they are
(34:30):
already involved in theecosystem and can kind of grasp
that more rich, platform story.
I look, there's a ton of thingswe didn't get to that I would
love to ask, but we are runningoutta time, so I wanna be
respectful of yours and theaudience.
I wanna wrap it out with acouple of rapid fire close out
questions.
You up for it.
Ryan Wang (34:46):
That's it.
Dan Balcauski (34:47):
Awesome.
Well.
When you think about all thespectacular people you've had a
chance to work with, is thereanyone who just pops to mind and
has had a disproportionateeffect on the way that you think
about building, runningcompanies?
Ryan Wang (34:59):
Yeah.
The person that comes to mind Ihas nothing to do with
technology, so a woman namedEmily Oster.
She now is well known forwriting a bunch of data-driven
books about parenting when I wasin high school and she was a
second year, I think, assistantprofessor at the University of
Chicago Department of Economics.
(35:19):
So.
Pretty busy time for a prettycompetitive place to be.
I emailed her and as well as abunch of other people at the
University of Chicago asking,Hey could you be a mentor on
this research project?
My, my school had this kookything where you could do
Wednesdays and you would go do aresearch project with somebody.
(35:41):
it, naturally most of theeconomists at, the Virgin
University of Chicago departmenteconomics were pretty busy and
did not respond.
And Emily did and she said, surething.
And come by my office and shemyself and my project partner
just sit outside her office andwe must have been 17 or 18 and
it, to this day, I still reflecton, wow, this person so busy.
(36:06):
So high powered, so much otherstuff going on, but took the
time outta the day to not justrespond to the email, but to
mentor us.
And so, there's that truism.
You send a good email and youfind that people are are pretty
generous with their time.
That was my experience and I tryto pay that forward a lot.
Dan Balcauski (36:20):
Well, props to Emily.
Yeah, it's amazing the impactyou can have on others when you
stop and give some of your time.
And there's also that truismthat like the busiest people are
the.
Easiest to reach.
It's like the,
Ryan Wang (36:31):
Right.
Dan Balcauski (36:32):
they there's either someone will answer email
in five minutes or they'll neverrespond.
And like the folks who answeredin five minutes also tend to be
the people who are like, I neverthought you would respond to
that email.
So props, props to Emily.
It made me think of, there's agreat book.
I've been in the consultingworld, a great book how to Talk
So Kids Will Listen.
It's a parenting book as well.
But it's awesome for aconsultant who's trying to speak
(36:54):
to clients because a lot of thesame principles apply.
And that's not to demean that'snot to mean clients in any way.
But oftentimes we don't make ourmessage heard.
In a way that can be heard bythe other party.
So, always a important lesson.
Uh, how do you stay sharp as aCEO?
What are you reading, listeningto learning from right now?
What any, uh, key sources thathave really changed your world
(37:15):
for you recently?
Ryan Wang (37:18):
Yeah I think two very different directions.
I very long ago.
When I was at the University ofChicago actually I took this
writing class and the writingclass was all about science
writing.
And because we were trying towrite like an honors paper or
something.
And the instructor, it was, Ithink it was called The Little
Red Schoolhouse.
(37:38):
It's a well-known class in abook.
And he was explaining actuallykind of similar to you, how do
you write for the audience?
And a lot of the times you writethis way because that's how you
think about it when you really,you have to explain it this way
'cause that's how the audienceis thinking about it.
anyway, he said the Nobel Prizeis an amazing example of this.
You go pull up the prizeannouncements for.
For physics, for biology, forchemistry, et cetera, et cetera.
(38:02):
And you find that there's threeversions of every announcement.
There's the public pressrelease, there's the technical
background, and then there's thespeech.
Same topic, very differentlevels.
Dan Balcauski (38:12):
Hmm.
Ryan Wang (38:13):
so I find that reading through these both as
interesting in an exercise ofhow do you dumb down concepts,
not dumb down, explain indifferent ways.
And just keeps, it's justinteresting.
It keeps me thinking about likewhat is happening in the world
that's coming not just nextyear, two years from now, but 10
years from now, 15 years fromnow, 20 years from now.
The total other direction fromthat is I love the 20 VC
(38:33):
podcast.
They have this group panel nowwith R Driscoll and Harry
Stabbings and Jason Lemkin, andthey just talk a lot about
current affairs and specificallywith this very deep lens on
technology.
It's very kind of turpentinetactical, which is what's useful
for me in the very, very nearterm.
Dan Balcauski (38:50):
Nice.
Nice.
Well, if I give you a billboard,you can put any advice on there.
For other B2B SaaS CEOs tryingto scale their companies, what
would it say?
Ryan Wang (38:58):
I.
I think I'd say, why are youpaying for billboards?
Get on the plane.
But seriously, I do think mybelief is that, you win
customers fundamentally by wordof mouth.
Yeah.
Awareness helps goose that forsure.
But at the end of the day,there's nothing that beats
finding somebody, having themtake a bet on you.
Doing a really good job for themand them telling everybody else
(39:20):
that they know, like, Hey, wow.
Assembled their AI agent ortheir workforce management
platform.
Here's what they did, and thenpercolate that out.
And then I found that all of thegreat companies started there is
do a really good job.
And then your best marketing isyour best awareness is the
people who you've already workedwith.
Dan Balcauski (39:39):
Do a great job and get on the plane.
Love it.
This has been fantastic, Ryan.
If our listeners want to connectwith you, learn more about
assembled, how can they do that?
Ryan Wang (39:46):
We're at Assembled.com.
We post a lot of stuff on ourLinkedIn, especially around the
intersection of AI and customersupport.
It's best place to check us out.
Dan Balcauski (39:54):
Awesome.
Well, I'll put those links inthe show notes for listeners.
Everyone that wraps up thisepisode of Sask Scaling Secret.
Thank you to Ryan for sharinghis journey and insights.
For listeners, you found Ryan'sinsights valuable.
Please leave a review and sharethis episode with your network.
Really helps the podcast grow.
Popular Podcasts
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.
The Bobby Bones Show
Listen to 'The Bobby Bones Show' by downloading the daily full replay.