Why Usage-Based Pricing Nearly Killed This 8-Figure SaaS

Omer Khan [00:00:00]:
Ryan, welcome to the show.

Ryan Wang [00:00:01]:
Thanks for having me.

Omer Khan [00:00:02]:
It's my pleasure. Do you have a favorite quote, something that inspires or motivates you?

Ryan Wang [00:00:07]:
Yeah. I love this. And I gave this at an all hands recently. This Robert Louis Stevenson quote. I have it in a doc next to Michael Jordan 1. It's don't judge each day by the harvest you reap, but by the seeds that you sow. And it was the start of fall when we did this all hands, so it felt particularly apt. But I do think as well, it's one in which you always look at today like, ah, did we hit this milestone, that we hit this goal? You know, what amazing things happened or what terrible things happened, you know, on the other end.

Ryan Wang [00:00:40]:
But it's really all downstream with things that you were doing a month ago or a quarter ago, a year ago, a lifetime ago. And I really like that quote in terms of just the groundedness.

Omer Khan [00:00:51]:
Yeah. And you know, people might be thinking, well, Ryan, it's okay for you. You've built this great startup and you think about seeds and all this stuff. But we'll talk a little bit about that in more detail because there was a time when you launched the business and it was a very tough time and you still lived by that same principle.

Ryan Wang [00:01:13]:
Yeah, absolutely.

Omer Khan [00:01:14]:
Cool. So for people who don't know Assemble, tell us, what does the product do, who's it for, and what's the main problem you're helping to solve?

Ryan Wang [00:01:22]:
Assemble is the AI platform for customer support. So we work with hundreds of modern enterprises, companies like Stripe, like Robin Hood, Etsy, Ashley Furniture, and we help them in a bunch of different ways. One is AI automation. So our chat agent, our voice agent, they handle 70, 80% tickets themselves. Our AI copilot makes people more productive, 20% more productive. It helps them ran 50% faster. And then our AI operations product, it's called workforce management. It powers the forecasting and staffing and orchestration between human and AI agents for contact centers as large as 20,000 people.

Ryan Wang [00:01:56]:
And then, you know, all the agents underneath, AI agents underneath. We've been at it since 2018. We launched in 2020. We were really slow to get off the blocks and we've raised 71 million in venture capital from NEA from Emergence Capital from Stripe.

Omer Khan [00:02:10]:
Awesome. And give us a sense of the size of the business. Like where are you in terms of revenue? Customers?

Ryan Wang [00:02:16]:
Yeah, we're at tens of millions in ARR and several hundred customers.

Omer Khan [00:02:21]:
And how big is the team today?

Ryan Wang [00:02:23]:
So we're about 122 people. So San Francisco, New York, and we just on Monday at our first person based in London start. So about half the company in San Francisco, about a quarter in New York, quarter remote and then one in London. And we'd like to grow that really quickly.

Omer Khan [00:02:40]:
Cool. So let's go back to when you launched the business. So it was just around the start of the pandemic. Where did the idea come from and why did you feel this was the thing that you were going to commit your time to?

Ryan Wang [00:02:58]:
Yeah, so it's odd. My own background was machine learning for fraud detection. So I joined Stripe back when it was around 80 people. And one of the striking things about stripe at 80 people among many was that Patrick and John, the co founders of Stripe, they would do support themselves more than that. They would have people over to their apartment to do all company support rotations. And so all across the company there was this ethos of that customer support really, really mattered. The customer experience really, really mattered. Lots of companies have something like this too.

Ryan Wang [00:03:40]:
When you peel back the early days, Slack had this, Zapier had this. Lots of iconic companies, but I was lucky enough to be there when it was going from 80 to doubling every single year. And obviously that's not going to scale. One of the first things as a machine learning engineer, two, three years into the company, remembering when the founders were doing support themselves is how can we point machine learning at this problem and help scale this ethos? From the early days, I started building tools with my now co founder, Brian Z, who was one of the first salespeople at Stripe. And we were trying to automate support, we were trying to categorize tickets, we were trying to do a lot of the things that AI does today with ML tools. But that wasn't really the problem. It wasn't really that they were looking to automate. I remember we took our system, it started as something called Agent Assistor, we renamed it called Lumos and people got a lot more interested after it was called Lumos.

Ryan Wang [00:04:40]:
But the head of support, Bob Van Winden, told us, you're automating 20, 30% tickets. But this isn't really the thing that we need help with actually. We're adding tons of headcount and we're hiring people across the world with different specializations to support our different products. There's all these layers of complexity and we actually need a problem with this thing called workforce management. So what's workforce management? He said, well, it's the forecasting a volume. How many emails, how many chats, how many phone calls. Are we going to get, we want to respond to people in a minute, in an hour, in a day, depending on the different queues and how many people need to be there and literally how many people are there, like the scheduling system for that. And we said, oh, you know what, that's actually a machine learning problem too.

Ryan Wang [00:05:26]:
You describe time series forecasting, you describe schedule optimization, constrained optimization, you describe some queuing theory in there and we're then off to the races. And a lot of this we were building for Stripe initially, but again, we had kind of really clearly observed, oh, you know what, it's not just us that has this problem at Stripe. There are a lot of companies that have similar kind of trajectories and so that naturally turned into Assembled.

Omer Khan [00:05:53]:
So was this a problem that fast growing startups like Stripe were having just because they were hiring so much headcount, or was this something that you were seeing with other companies that maybe had a large support team but weren't necessarily growing as fast?

Ryan Wang [00:06:10]:
Yeah, I think at the time we didn't understand, but a lot of it did have to do with growth. So early on, Bob, we realized had come from Google. And so Google had this workforce management problem with some particular acuteness. And Slack as well was one of our early customers. And yes, they were growing really quickly. And then we I think later happened upon, oh, you know, this is a huge category. And you know, bank of America and JP Morgan and United Airlines and all these kind of iconic brands, iconic consumer brands, they have like a different version of workforce management that is much, much more complicated. And it got installed years and years ago, even decades ago.

Ryan Wang [00:06:57]:
You know, it looks like Windows 95, but that came later. Our first segment really was fast growing startups and that kept us busy for quite some time actually.

Omer Khan [00:07:10]:
So tell me how you went from building an internal tool for the Stripe support team to Ryan and Brian, co founders of Assembled.

Ryan Wang [00:07:21]:
It happened very gradually. Then all of a sudden I think my brother, who's our third co founder, he was getting ready to wind down a startup that he had done with Y Combinator with some of his college friends. Brian had gone off to Southeast Asia to rediscover himself and he was coming back and then we were all trading ideas and we were actually in Bozeman, Montana. So we were in Montana before. It was cool. This was like 2018 and we got together and we were just hacking around with some ideas and we realized, oh, I think we know what it is we're trying to build. John had written on a piece of paper Kind of the prototype of nowadays you would vibe code it, but he just kind of drew it out, the design of the first product. And we said, you know what, this makes a lot of sense.

Ryan Wang [00:08:17]:
And then at the very top level, we feel like we have this thesis about customer support. And the thesis was really as simple as most companies want to scale great customer support. They want to have good customer support. As a consumer's writing in, I think we've all been there, having that moment where it's, what the hell is going on here? Do these people care at all? But inside the company to a T, every single company we met, every single company we've worked with over the years, now they're trying to get it right. It's just a really hard operational problem. And so we had the thesis that companies wanted to scale great customer support. We had a piece of paper with a picture and we said, you know what, let's go do it. And you know, that question really has persisted, I think, throughout the company.

Ryan Wang [00:09:04]:
All of our products, AI, not AI, you know, automation, augmentation, operations, not operations. It's always been in service of how do you scale great customer support? So I think once we had clarity on, oh, yeah, that's the question that we're trying to answer. And we really knew this should be a company.

Omer Khan [00:09:23]:
So your first customer was Stripe. And in many ways you built the product for Stripe. Right? You were given the requirements for what needed to be solved. And when it came to getting the next few customers, how easy or hard was it to convince them that they needed a solution like this?

Ryan Wang [00:09:50]:
Yeah, I would say it wasn't so much convincing them they needed a solution as discovering their problems and then realizing, oh, wait, we can draw a line between them and kind of almost like regression style, fit the line to the curve or fit the line to the scatter plot. And what I mean by that is Stripe was our first customer, probably much larger than everybody else. Our second customer was Casper. And I think we gotten introduced to them by an investor that was probably trying to vet us a little bit. And there was this guy, Gabe, who was the CTO of Casper. And he had this incredible philosophy back then of bet on tools. If they bet on the right early companies and the right early tools, they would get cheaper software and be able to benefit by pointing people at the roadmap. So Casper was a lot smaller than Stripe, but growing really quickly.

Ryan Wang [00:10:49]:
And then our third customer was Grammarly, and that was a friend of a friend who is a PM there. And he introduced us to the Support team. And then our fourth customer is GoFundMe. And. And each of them, what we found was, what are some of the challenges that you have in customer support? What does it look like? We're just following them around, following them around with all four of them. There was a spreadsheet with dates and times across the top. So you're starting to imagine a calendar and the names going down and then this color coded. What are people supposed to be working on? And everybody had different colors.

Ryan Wang [00:11:29]:
And the day started a different day of the week. Some started on Sunday, some started on Monday. And we literally learned because grandmother was in Ukraine. And so it's a different start of the week, but you were like, oh, my gosh, you all have done the same thing, different solutions, but almost. Almost exactly the same to the same problem. And so across these different points of scale, we drew in the line between. And we felt like, okay, we can fix that for you and we can make this problem go away. Because the people on the support team that had to use it, they had to pull up the spreadsheet, and at larger and larger scales, it just wouldn't load.

Ryan Wang [00:12:17]:
Google Sheet with a lot of data in it. And you had to go find your name and go search through it. It's like, oh, ye, that's what I'm supposed to be working on today. That's when I take my break today. And then if you were the manager of support, it was one person. So you're like, oh, you're managing hundreds of people and this one person is responsible for this tool and it's very brittle, so if it breaks, you're in big trouble. So, yeah, it wasn't so much convincing them as discovering, oh, you all have the same problem. We could put this into software.

Omer Khan [00:12:45]:
I often talk to founders, heard it multiple times, that whatever niche or market they're going after, the customers are currently using some kind of spreadsheet and they're building software to help them replace the spreadsheet. But often what they find is that people talk about this manual process and it's painful, but at the end of the day, it's not really that painful. And when you go there and try to sell something, they're like, it's not that easy to sell the product or convince them. So I'm curious, what was it about these specific examples where you figured out there was enough pain there that people would care enough to do it differently?

Ryan Wang [00:13:34]:
Yeah, and I agree. I think even to this day, if you're below a certain team size or you're below a certain complexity on that workforce management product. It, it's probably not going to be the thing for you. You probably are better off with spreadsheets, so you can't hang your entire hat on. We're replacing this process. I think for us, what we discovered organically was that that was actually kind of the foundational data, if you will, for a lot of other really interesting things and a lot of other valuable things within customer support, where it wasn't just, hey, we're taking away the manual process of the manager or even just this kind of relatively small pane of what support agents had to look up, but rather within Casper, for example, we were walking around and we saw these TV dashboards in their office. It sounds a thing of the past now, but they had TV dashboards because everybody was in the office. And the TV dashboard showed the state of the queue and what everyone was up to.

Ryan Wang [00:14:42]:
And we later found out, hey, if we know what people their schedules are and then we can kind of connect that to the state of the data from your Zendesk or from your telephony system, from your Salesforce service cloud, we can show you this real time dashboard of everything that's happening. And, oh, if you've got customers waiting and you've got people on this other queue that are available, you can move things around. And that also translates to the AI world. So that same kind of concept. Now, while some of this is human capacity, some of this is AI capacity. You can dial it up and down, but all of that kind of capability is powered by, well, we know what people are up to and we know what they're supposed to be doing. And we know who's available today, next week, next month. There's another piece where, well, you want to forecast it out and everybody wants to get their hiring plan.

Ryan Wang [00:15:36]:
It's such a pain in the butt and it's a huge, huge cost. Like a lot of companies, we're talking tens of millions of dollars a year spent on customer support. So getting the headcount planning right is really quite a large financial exercise. And again, you want to know about productivity, you want to know about attrition rates, you want to know about utilization rates. And that comes out of the schedule as well. Right. And that's a hugely valuable problem. And so we added on all these pieces around the schedule.

Ryan Wang [00:16:10]:
I think early on we started with this very narrow pane and then over time discovered these other pieces that we could add onto it. And then eventually we ran into the kind of existing category. There's A story past that, too, where workforce management has been a category for a long time. I mentioned, and one of our customers during the pandemic was Robinhood, and they just absolutely blew up. I think they started as maybe 100, a couple hundred people, and there was a point in time between Bitcoin and meme stocks and their whole C suite doing customer support because it was so backlogged to the headcount, just went vertical. And that was amazing for our business at the time, but also incredibly stressful because you start to bring in people from Charles Schwab, you start to bring in people with real enterprise grade experience, and they're looking, looking around, being like, what is this assembled piece of crap? This is like a little toy. We need a real enterprise grade system. We're like, no, no, no, we can get there.

Ryan Wang [00:17:07]:
Just give us like a month. And so we were just coding furiously and then also importantly, getting up to enterprise grade. And so that kind of experience in lightspeed ripped us all the way through to. Okay, it started as a spreadsheet replacement. We're adding on all these value add features. And then at the end of it, just at the end of this gauntlet. Yeah, now we can replace enterprise tools that have been around for decades.

Omer Khan [00:17:30]:
Yeah. I mean, I think the insight there for me is like, it started as a spreadsheet replacement, but if that's all you're doing, it's a really superficial solution. Right. You're just taking a spreadsheet from a spreadsheet and putting it into software. And it's much more about the reason you were able to get traction was because you understood the nuances and the details behind that and why they were doing things that they were doing in the spreadsheets and what were the problems behind the spreadsheets and so on, that was a much bigger thing to go and solve. I'm curious, you kind of had sort of a plan for this sort of biggish launch and getting featured on TechCrunch and all of this stuff. And unfortunately, it was the same time as the pandemic hit. So can you just tell us what happened?

Ryan Wang [00:18:26]:
Yeah. So we had raised the seed round, and one of our friends, who is Kelly Simmons, she was the first PR person at Stripe, had told us, nobody writes about seed rounds in TechCrunch. We're like, okay, well, that stinks. But we're still trying to figure out a moment to launch. And then she said, but I know this person, and if you figure out how to hustle in front of her, and she's in London and I know somebody who's going to be in London. Maybe she'll write about you if you kind of intrigue her enough. And so we really pulled some strings to get this kind of story lined up for it was a $3 million seed round. Nowadays it's even worse, right? I can't imagine anybody writing about a $3 million seed round.

Ryan Wang [00:19:18]:
But we had this TechCrunch article ready to go. And the day kept moving because we were talking over WhatsApp like, hey, is the article ready? Oh, no, I just need to touch it up one more time. Ah, shoot, we got to be ready to do this tomorrow then. And then it was. So this was March 2020, and everybody already kind of had inklings of what was happening. And then I just remember, okay, today's the day, and telling the team, let's all go into the office. And we had this small little live work loft and there are probably six or seven people at that time. We went into the loft and I think everybody was a little bit like, is this okay? It's probably fine.

Ryan Wang [00:20:03]:
None of the stay at home orders had come out yet, so it was probably fine. And then we had also worked really hard to get our post onto Hacker News. So we had people kind of ready to share about, upvote and kind of comment on our Hacker News launch post. And it worked. So that day we got to the front page of Hacker News and I still have the screenshot on my computer of World Health Organization declares Covid a global pandemic. Assembled launches. And then World Health Organization declares Covid a global pandemic. And so we're here in this apartment like woo hoo.

Ryan Wang [00:20:47]:
And high fiving like, wait a second, maybe we should all just leave right now. Is this safe? And you know, I think the day after, the week after we realized, you know, it started to settle in and all this stuff was happening. Oh, you know, this could be really bad. You know, we had all this interest. I'm from Chicago and I'm a huge Cubs fan. Just, you know, grew up a Cubs fan. And somebody from the Chicago Cubs wrote in to, you know, get a demo of assembled and we're so excited. And he never showed up.

Ryan Wang [00:21:21]:
And that happened. Probably the 25% of our demo requests were, shoot, do we just waste this launch moment that we worked so hard to put together? And of course, a month later you come to find it was super bad, very briefly. And then it was amazing. And then all of the things that people companies scaling, there were so many companies Scaling so fast that needed help with headcount planning, that had to add support, support, support, and couldn't get their hands around it. So ended up being really, really good for our business, despite that kind of shocking moment.

Omer Khan [00:21:57]:
So let's talk about the period where it wasn't good for your business. So the reality settles in. And then there's this sort of time where you said a lot of people aren't turning up for these meetings and many companies just stopped buying anything. Right. So when that happened, what did you decide to do? Where did you decide to focus your energy?

Ryan Wang [00:22:22]:
You know, we were too. I don't know, maybe it was naive to understand that it was macro related because at that point in time it was a SaaS product, but we had just come out of stripe, so we only knew usage based and none of us really understood SaaS. Maybe if we listened to this podcast we would have understood, hey, have some contracts have a minimum. But it was just, hey, you sign a piece of paper with us that tells you the price, but you could turn it up, you could turn it down. Totally up to you. So we felt like we had to go really earn every single bit of usage. And that usage was, if we didn't have that, we wouldn't bill. And so we were kind of in there with people with those very early customers, trying to convince them to add more people to, to the system and trying to show them value so that they would keep scheduling people ultimately was the unit of billing.

Ryan Wang [00:23:27]:
And then all these other kind of value adds come out of the scheduling data. And so, yeah, we definitely saw that flatline shrink and we thought it was our fault. We thought, oh my gosh, something between the last couple months, like, what are we doing wrong? Are we, are we not paying enough attention in onboarding? Are we missing some killer features? It didn't occur to us until a couple months into it that, oh, maybe there's this macro thing happening. So we really beat ourselves up over it.

Omer Khan [00:24:00]:
And I think from what we were talking about earlier, you sort of shifted from trying to scale this business to focusing much more on like one customer at a time. Which I guess connects back to the quote that you shared at the beginning, which was, hey, there's no harvest in sight, so let's start planting some seeds, I guess. Tell me about that.

Ryan Wang [00:24:30]:
Yeah, 100%. I think we just focused in on the people that we knew were getting value and kept building around them. And, you know, I, I think we had this hunch that if we got the platform to a certain kind of critical mass, then, then people would use it. We just need to get over that hump. And you know, even it goes earlier back, I remember it was very hard for us to get our first dollar of revenue. And it was, like I mentioned, it took us a long time to launch. So 2020 was two years after we theoretically started. So we were kind of used to this feeling that maybe things will be harder until you reach that point of critical mass.

Ryan Wang [00:25:19]:
And I talked to Jack Altman really early on. He was an angel investor and he had started this company, Lattice. He's obviously a famous last name. And he said something really that had stuck with me over the years. He said he had gone to a really good Ivy League school and he worked at a pretty successful startup and rose super quickly. And then he and his co founder, same type of backgrounds, went to start Lattice. And then the way he described it was success was not linear. It didn't matter how hard you worked or how creative you were, you were just going to have to hit your head on the wall for a little bit.

Ryan Wang [00:26:02]:
But everything in your career up to that point had been if you work harder, if you are more creative, you will go far. Right? He said it took them eight months to get their first dollar of revenue. And I think we experienced something very, very similar. And he said that just shook off a lot of the feeling that, okay, this is a linear path. Instead, we're going to have to take some bets. We're going to have to have some conviction on where to double down. We're going to have to squint a little bit to see what's working. And I think for that period, for us, it very much influenced that mentality of, okay, there are some features that we know you just have to work like schedule generation.

Ryan Wang [00:26:45]:
There's something where you can kind of click a button and generate this optimal schedule. And it's kind of an enterprise type of feature and it's pretty complicated and there's some algorithmic stuff underneath. You can't iterate your way to that. You're just going to have to get it right. And until you get it right, people are going to. They're not going to be able to really see what you see. And that's just going to take a while. That's a hard thing to do.

Ryan Wang [00:27:11]:
So we had a couple features like that where we just knew we had to buckle down. And then also there were some features where, yeah, it was like, you're telling us this stripe, but we don't know if that's really Generalizable. And so there was a lot of conversation, too, around, okay, we're trying to wrap our heads around this problem more deeply. Wrap our heads around this problem more deeply. And that will help us a year, two years, three years down the line. I don't think we really understood in that moment what we were doing, but we were really collecting data to have this mental model of how do all different sorts of organizations handle this problem and related problems in customer support. And that ended up being hugely valuable. So even if we didn't see see the output of that, the harvest of that immediately during COVID it did end up being very useful down the line.

Omer Khan [00:28:03]:
So one of the traps that founders sometimes fall into when they're working closely with one customer at a time is basically building a custom solution for each customer. Did you guys ever fall into that trap?

Ryan Wang [00:28:19]:
Definitely, we. And we would continually introspect that question because we were engineers and Brian was kind of sales. John is kind of classic engineer. I was kind of in between, but kind of my most recent professional experience had been engineering. So we had this really dumb argument. Not even dumb, but just not even argument. When we were signing Robinhood, you know, Brian kind of showed up and he's like, guys, I have this multi 100k deal with Robinhood, and John and me and Michael was one of the early engineers, and Lyn was one of our early engineering teams. We're doing a lot of custom stuff already for other customers.

Ryan Wang [00:29:09]:
Maybe we don't want to do this with Robinhood. Brian was CEO at the time then. So now if I were, I think, in that seat, I'd be like, what the hell are you talking about? Do you think 100k deals, do things grow on trees? What are you talking about? Figure it out. But yeah, we thought, hey, we're doing too much custom work. And we really sat down as a group and kind of peeled apart. Okay, here are the things that we're doing for Stripe that are custom. Here are the things that we're going to have to do for Robinhood that are customers and which of these generalize? And we wrote a really long doc to convince ourselves that we should indeed work with Robinhood and not walk away from the deal. We still have that doc, and I show it to people sometimes.

Ryan Wang [00:29:56]:
I think that is the right way to think about it. You sometimes have these deals, and, yeah, it can hamstring your company if it's the wrong one. And it can be. You need the blueprint for the next stage if it's the right one. And Our experience has been that if you do get the right customer, like if you build around the right customers, and my mental model is still, you're kind of like drawing out the line. And so great. You have this line built up to the size of your current customers and you have somebody that stretches you all the way out. And your question is, does that generalize or are they a total weird outlier? And if it generalizes then that's going to be the beachhead to the next set of multi hundred k customers.

Ryan Wang [00:30:40]:
And that indeed was the case for us. Robinhood was an amazing customer and we were so glad that we built to spec for them because it did generalize. And then our first million dollar a year customer, it was the same type of thing that was the first million dollar customer. All the custom things ended up being the things that other million dollar customers would need. And then we do have a couple examples of that exercise where we walked away from deals. Like there was a big airline that we thought about working with who was on like a Microsoft Dynamics CRM and so we would have to do that. And we had slack notifications built into our system and they would have needed us to move that to Outlook and to teams. And so we walked away.

Ryan Wang [00:31:25]:
And I think that was the right move because we were not ready for that ecosystem that would have been just purely custom and we would have been stuck for a little bit. I do think that the move from custom to configurable, that's a difficult one to get right. But it really does require the founder to apply a lot of judgment or the team to apply a lot of judgment around. Does this generalize?

Omer Khan [00:31:50]:
So let me just recap a bit. So it took you quite a while to get started with these kind of, you know, the launch that didn't actually work out to be much of a launch. It took a long time to get that first dollar. Things suddenly came to standstill because of the pandemic. You're working, you know, with one customer at a time. This is sort of a custom component being built for, for each customer. And then there was this one year where you went from like 10 customers to 50 customers because you, you were focusing on the shift to more configurable and you sort of blew past the first million in ARR during that time. I'm curious, did anything break with all the custom stuff that you had built while you're suddenly ramping up to like 5 times the number of customers that year?

Ryan Wang [00:32:42]:
Yeah, definitely. I think we had you mentioned it was super hands on and Even to this day, we have a cultural value called get on the plane. So we were meeting customers in person. We were doing onboarding ourselves. We were highly incentivized to make sure they adopted because otherwise we would not get paid. And that only goes so far. It works for 10 customers. You can get on the plane.

Ryan Wang [00:33:10]:
We even went to Kyiv in 2019. So that's a special place for me. But, yeah, that's a long flight. That's a long several series of flights. And we were going routinely back to San Diego for one of our other customers. And every single week, John and Brian would get on a plane and go to San Diego to the point where they had a favorite Chinese spot that they would go with this team. So, yeah, that works at 10, doesn't work at 50. And that was around the time when we brought in our first designer and she was a contractor, and we really wanted to hire her.

Ryan Wang [00:33:46]:
She didn't ultimately join us. It's a different story. But what we asked her to do was take a magnifying glass to our onboarding process and figure out what are all the pieces that are super manual and we can automate. And the goal was just take something that takes weeks and takes a bunch of effort and compress it into days and figure out how to have it be automated. So onboarding broke, and we fixed it by pointing design at it. Definitely something that broke was in the early days, it was super easy to ship a bunch of features that maybe were half baked. And I think especially in enterprise software, there's a big difference between the demo and the thing in production. And I think this is one of the fundamental failures of how to purchase software.

Ryan Wang [00:34:36]:
There is a bit of a mismatch between what people are looking for at that point point of buying and what actually happens. So in truth, we had some features that we had built for the demo, and there was nothing there that didn't work. And so some of them, we had to say, well, now it's time to make that thing work. That we showed people. And some of them it was, you got to kind of step around this corner, just know if you're doing the demo, that thing does not work. And so it was probably, we should get rid of some of those. And we had a lot of cleanup work to do and engineering parlance, a lot of technical debt to clean up so that people coming in would come in like new hires would be, what the hell is going on with the configure staffing settings? Oh, yeah, don't worry, that never worked. So you're not crazy.

Ryan Wang [00:35:27]:
That did not work.

Omer Khan [00:35:30]:
I want to talk a little bit about just growth. And I know you mentioned most of the early customers came from word of mouth, warm intros, and once you scale beyond a million in ARR, you started to build out your GTM motion hiring AES and all that stuff. But communities played a big part in your growth as well. I'd love to understand. Like, you know, we sort of know the playbook in some sense, right? It's like, go and find where your customers hang out, the communities and. But turning it into something that helps people understand about your product without being pushy or salesy or annoying is kind of the balance. Right. So what did you do? What were some of the lessons you learned from being part of these communities?

Ryan Wang [00:36:24]:
Yeah, I think the big lesson there was whether it's community, when people talk about gtm, they talk about kind of the lever or the motion community or outbound or billboards or content like nurture. We had read through all these, and I was interim head of marketing for a while before we hired a head of marketing and kind of learning all this stuff. And a lot of the literature was about levers. And I think what we realized was, and it was by accident, one of our customers from Casper was texting these other kind of customer support leaders in the New York startup scene. And we heard about it and it's like, oh, how did you come to know each other? And they told us about this slack community called support Driven. Okay, well, we have to join this slack community. And you join and you're like, whoa, everybody's here. This is where all people are hanging out.

Ryan Wang [00:37:32]:
And people would routinely trade notes there. How did you make your first hire? Hey, we've got 10 hires. We're trying to figure out how to scale up. How do you think about outsourcing? How do you think about workforce management? And we knew things were working when people started to ask, hey, we're looking at this tool called Assembled. What do people here think about Assembled? And then, you know, it. It occurred to us, well, we should kind of grease the wheels a little bit there. And you know, in the early days we would, we would ask people, hey, there's this thread, you know, that that's about assembled. You know, would you be up for just sharing your thoughts? Like, you don't have to say what you don't really think, but would you be up for sharing your thoughts? And to a T, you know, because we were super hands on with people at the time.

Ryan Wang [00:38:19]:
They would, oh, yeah, of course. Love you all. Happy to share my thoughts. And then you'd start to see these threads where what do people think about Assembled? Assembled's amazing. Oh yeah, I also had a great experience. Oh, it's super good for replacing your spreadsheets. Here's how you talk about your finance team. Right.

Ryan Wang [00:38:37]:
So do, do, do, do, do. And then we quickly found that basically we had everybody that was looking for software in that audience, in that community was essentially team assembled. And I think the learning from there was, it was less about we went looking for a community and it was more where are our customers hanging out and in particular, what is this audience? And I think about it now with our team as we're trying to get to the same clarity with some of our new products. What's working and where are customers super happy? And where do you see three, not one? And what are the commonalities and just where do they hang out? And so think about that as what ponds should you be fishing in? Because once you've kind of gotten full mind share of that one, then you should think about kind of leaping from there to the next audience. But you see that these are kind of pockets of kind of like densely connected people. And that's what we're really trying to do across all these levers is figure out how to win that audience.

Omer Khan [00:39:39]:
Yeah, love that. Before we wrap up, just one last question about your icp. Typically we know have a specific icp, all of that stuff, but when you actually ask many founders at the early stage, who's your icp? It's kind of very fuzzy and vague. You were kind of very data driven about your icp. Just tell me a little bit about what that involved. And more importantly, how did you figure out that level of detail?

Ryan Wang [00:40:10]:
Yeah, our first go to market hire was this woman, Jen Ong Vuong, and she came from Stripe with us. And it was her, I would say, scrappiness that. And we take this as a playbook today in our early workforce management product, in our AI products, it's just sit down with the list of customers and play around with it. And so even in our most recent board deck, it was, hey, give us all of our AI chat customers, give us all of our AI voice customers, and give us as many columns with characteristics there. And we're just going to look and stare at that list because a lot of the times I think when you try to do the ICP exercise, it's, you know, kind of thinking or shooting from the hip or trying to be strategic. And what Jen did was just take our list of customers and, you know, kind of like sticky note them together. Oh, like, these ones are all on Zendesk. They're all on the same platform.

Ryan Wang [00:41:05]:
Oh, these ones are all on, you know, they're all direct to consumer. It's direct to consumer. These ones, they're. This one's 20 agents, this one's 30 agents, this one's 10 agents. Oh, there's like a grouping right here. And so you had these groups of Tier 1, Tier 2, Tier 3, across all these different characteristics. And then it wasn't as clean as that. I think it took a little bit more time, but that was the process.

Ryan Wang [00:41:30]:
And then by the end of it, it was, oh, our first set of workforce management customers, they were 20 to 200 support agents. They were on Zendesk or Intercom or Customer. They had support hours that were not just nine to five, but like seven days a week or cross time zones so that it doesn't fit one shift. But you have to have multiple teams, they're on multiple channels. It's not just email support. You just kind of show up and do as many emails as you can, but you have phone support, so you need to know, okay, we have to have people there or chat support, Same type of thing. And so it was as specific as that, but it fell out of, okay, what is going on with simply the customers that we have if we try to look at really hard at them.

Omer Khan [00:42:14]:
Yeah, yeah, I love that. We should wrap up. So let's get onto the Lightning round. I've got seven quick fire questions for you.

Ryan Wang [00:42:23]:
Sure, sounds great.

Omer Khan [00:42:24]:
Okay, what's one of the best pieces of business advice you've received?

Ryan Wang [00:42:28]:
We worked with an executive search firm that said this deeply annoying thing, but that I think contained a lot of wisdom. They said, you're not stripe. We had come out of Stripe, and it was like, yeah, of course we're not stripe. We're assembled. But the way that we were operating was like stripe. And we needed to operate. Assembled.

Omer Khan [00:42:50]:
Love that. What book would you recommend to our audience and why?

Ryan Wang [00:42:53]:
I really like this book, and it's really all the books by this author named Steven Berlin Johnson. He had a book called Where Good Ideas Come from that was highly influential to me. And basically the idea of the book is it's not this amazing moment of inspiration. We're used to the kind of scientist who's in their house and then Eureka. But actually it's this very highly networked thing and it comes from people trading ideas. And there's this concept of the adjacent possible. And when multiple innovations happen across disparate fields and then they come together, the adjacent possible moves and multiple people get to the next insight pretty pretty much at the same time. And I think the takeaway for startups there is networking actually is a real thing.

Ryan Wang [00:43:41]:
You can learn a lot and you can get a lot out of being plugged into networks.

Omer Khan [00:43:45]:
What's one attribute or characteristic in your mind of a successful founder?

Ryan Wang [00:43:50]:
Jensen has that suffering thing, and I think that's a real thing. I think I was captain of a baseball team that went 0 and 21, and I think that was incredibly valuable for me. And I always feel a little bit like, ooh, should I share this? Will people think I'm a loser? I'm not a winner? And I do think it's important for people to know what does winning look like? But if you go through something where you're 0 and 21 or you're just kind of getting whooped, and every single day I feel like, okay, but the next one, if these things happen, we could do it. I think that's a good trait to have. I think that's a trait that you need to have, frankly.

Omer Khan [00:44:30]:
What's your favorite personal productivity tool or habit?

Ryan Wang [00:44:33]:
I have been using a tool called RescueTime for now, 20 years now. Oh my gosh, I wish they would go back to the. If anybody from RescueTime is listening to this, I wish they would go back to the older ui. But it just tracks everything you're doing on your computer. You can log offline time. It tells you at the end of the week, hey, here's how much time you spent across different tools. And mainly at this point, it's just useful. People talk about nine, nine, six.

Ryan Wang [00:45:01]:
I'm like, I know how much time I can work. It's about 60, 70 hours on the computer. And then if I go too far past it, I feel really tired the next week, and you can tell why. And if I go below it, then there's some weeks where you go below it and you're spending a lot of time in person, or you're just kind of walking around, or you're just resting and recovering. But it kind of gives you a baseline for, okay, here's roughly how much I can do.

Omer Khan [00:45:25]:
What's a new or crazy business idea you'd love to pursue if you had the time?

Ryan Wang [00:45:29]:
It turns out this is not just unique to me. I was talking to a friend, Ravi, from Heap, about this. I have a long neck, and so I have these neck problems. I would Love to have a computer desk where you can kind of lie down flat and just kind of stare at the computer above you and then you're basically a lying down desk. Apparently Alan Greenspan used to take meetings lying down. So maybe there is a market for it.

Omer Khan [00:46:02]:
Yeah. What's an interesting or fun fact about you that most people don't know?

Ryan Wang [00:46:09]:
I think if I weren't doing startups, I would want to be an economics or a statistics professor. I don't know if I have what it takes today. I probably don't have the attention or the focus to be able to make it all the way through a PhD program. But that's my lifelong goal, is to be a professor.

Omer Khan [00:46:29]:
Cool. And what's one of your most important passions outside of your work?

Ryan Wang [00:46:34]:
So I went to a boarding math and science magnet public school and so I really am interested in public education and in different creative forms of public education. I think it is the great equalizer. And yeah, this school is called the Illinois Math and Science Academy. And it was at a time in the late 80s, early 90s where the country, the world was really interested in how do we incentivize new creative minds? And I worry that we're kind of getting away from that. So I think public education is one of the great equalizers. And I think that there are a lot of things that there's still a lot of interesting ways in which people are pursuing this, but there should be a lot more.

Omer Khan [00:47:26]:
Awesome. Well, Ryan, thank you so much for joining me. It's been a pleasure. I think we covered a ton of really interesting stuff in the last, what is it, 48 minutes or so. So I appreciate you sharing your lessons and kind of the story of building Assembled. If people want to check out Assembled, they can go to assembled.com and if folks want to get in touch with you, what's the best way for them to do that?

Ryan Wang [00:47:55]:
I'm just ryan Wang on LinkedIn and you know, I look at my messages and share some stuff about the future of work and the intersection to customer support and so please give me a shout.

Omer Khan [00:48:06]:
Awesome. Thanks, man. It's been a pleasure and I wish you and the team the best of success.

Ryan Wang [00:48:10]:
Thank you.

Omer Khan [00:48:11]:
My pleasure. Cheers.