AI For Superhuman Customer Support | Ryan Wang

Hey CX Nation,

In this week's episode of The CXChronicles Podcast #278, we welcomed Ryan Wang, Co-Founder & CEO of Assembled based in San Francisco, CA.

Industry leaders like Etsy, Robinhood, and Stripe trust Assembled to provide customer-facing AI agents and workforce planning at scale. 

Assembled automatically resolves millions of interactions through chat, email, and phone while optimizing staffing for hundreds of thousands of support professionals. 

Their mission is to elevate customer support through AI-powered software that makes life easier for customers and employees.

In this episode, Ryan and Adrian chat through the Four CX Pillars: Team, Tools, Process & Feedback. Plus share some of the ideas that his team think through on a daily basis to build world class customer experiences.

**Episode #278 Highlight Reel:**

1. Building a high-performing team in the AI age
2. Shift towards AI-driven skill sets in the workforce
3. Creating a culture of continuous learning 
4. Focusing on customer feedback early & often
5. Keeping your team lean & flexible as you scale

Click here to learn more about Ryan Wang

Click here to learn more about Assembled

Huge thanks to Ryan for coming on The CXChronicles Podcast and featuring his work and efforts in pushing the customer experience & contact center space into the future. 

For all of our Apple & Spotify podcast listener friends, make sure you are following CXC & please leave a 5 star review so we can find new members of the "CX Nation". 

You know what would be even better?

Go tell your friends or teammates about CXC's custom content, strategic partner solutions (Hubspot, Intercom, & Freshworks) & On-Demand services & invite them to join the CX Nation, a community of 15K+ customer focused business leaders!

Want to see how your customer experience compares to the world's top-performing customer focused companies? 

Thanks to all of you for being apart of the "CX Nation" and helping customer focused business leaders across the world make happiness a habit!

Transcript

Adrian Brady-Cesana (00:00.943)
Alright guys, thanks so much for listening to another episode of the CX Chronicles podcast. I am your host, Adrian Brady-Cesana, and this is another episode, guys, that I'm super duper excited to share with y'all. We've got an awesome guest today. I know I say that all the time, but today's guest is gonna be a lot of fun. Ryan Wang, CEO at Assembled, is joining us. Ryan, say hello to the CX Nation, my friend.

Ryan Wang (00:21.134)
Hey, it's really good to be here.

Adrian Brady-Cesana (00:24.051)
100 % so guys number one. I just was joking with Ryan earlier this week We were already in a rock and roll and we are already to jump into this and then the wonderful joys of technology guys my my internet was not allowing or supporting for Ryan and I to jump in and to do today's conversation recorded, but we're back today and everything's good to go so right Thank you for being patient with me and Maria and our team here at CXC But we're pumped man were pumped to dive in today and we're really pumped to kind of share the story that That that you that you started to share with us last week

Why don't you start off today's episode, before we even get into the pillars, before we get into some of that, take a minute or two, to tell the CX station. Number one, introduce yourself, and number two, a couple of stepping stones. Spend a minute or two kind of talking about how did you get into this whole world that you're in today as the CEO of Assembled, and how did the idea of building Assembled even come about, man? I'd love for you to kind of start off with the stepping stones question.

Ryan Wang (01:15.564)
Yeah, for sure. And just really quickly on Assembled, we are the AI platform for customer support. So we're helping with all the different pieces of helping companies deliver great support with AI, whether that's customer facing chat agents or voice agents that are automating interactions, whether that's co-pilots that are making human agents more productive and drafting replies for them and doing translations, or whether it's all the way in the back office that I know a lot of folks.

here know, but maybe the broader world doesn't know around workforce management and forecasting and scheduling big teams that are distributed all across the world. So we're solving kind of all the different pieces of support operations working with companies like DoorDash, like Stripe, like Ashley Furniture. companies of all different shapes and sizes. You know, I think the way that we got onto this, my background is machine learning. So I was a machine learning engineer. worked on a lot of cookie data science problems at this

My first job out of college, I was at a consulting company. It was called Freakonomics Consulting. And it was one of the co-authors of Freakonomics. And then he roped in a guy named Danny Kahneman, who was the only psychologist ever won the Nobel Prize in economics. And he was doing a lot of data science consulting. Turned into a machine learning engineer at Stripe, working on fraud detection when it was roughly an 80 person company.

And this is roughly where my journey kind of intersects into assembled where in the early days of Stripe, when it was an 80 person company, the co-founders, they were doing customer support themselves and they were having the whole company, you know, over to their apartment to do customer support. Like it was a full company support rotation.

Adrian Brady-Cesana (02:57.651)
That must have been fun. That must have some good team building time right there.

Ryan Wang (03:03.734)
It was super fun and it was super intoxicating and it was super, for me, somebody who was working on fraud detection algorithms to then be pulled in front of a ticket where somebody said, hey, I'm seeing a lot of fraud or I had these charges declined, what the heck is going on? To try to be able to put together the answers to that, even knowing that I had worked on those models and then jumping through the different tools and trying to figure out, okay, well, how am I gonna find this? I gotta look at their data. It was really challenging and it was really fun.

And it was a great feeling to be able to say, okay, you we've been building this thing over here and now we're able to answer like the real question that people are grappling with over there. So I always kind of had this fondness for support rotation. then Stripe was a company that grew really quickly. So it went from 80 to a couple hundred, a couple thousand in just a handful of years. And when you looked at the company at a couple thousand, know, millions of customers, a lot of different products, you know, they started to bring in.

Adrian Brady-Cesana (03:51.28)
Absolutely,

Ryan Wang (04:01.054)
you know, different types of BPO's and so just like the whole thing got a lot more complicated. the question was for, for, for myself and my co-founders at Stripe, how do we draw the line between this huge problem at scale? And then when it was the co-founders and the whole company doing rotation in their apartment, right? So, so, so, so, you know, I mentioned we help companies deliver great support and we're using AI to do it. It really just as simple as that. Like, how do you draw the line between companies when they've become really big and really successful?

and that earlier ethos of wine to deliver great support.

Adrian Brady-Cesana (04:31.923)
I love, so a couple things to call it. Number one.

I don't think you told me about the Freakonomics connection earlier, like in our last, that's super duper cool. So folks who don't know, Freakonomics, that was a super, super popular book in the early 2000s. Right, I'm gonna have to pull this out. I literally have that book sitting on a shelf somewhere inside, so that's awesome. I mean, you just didn't graze upon that the other day when we were catching up on things, but super cool starting. then, you know, as you talk about just the machine learning background, and then you talk about having this incredible experience, because you did, we talked a lot about your experience at Stripe. Guys, for many of you who haven't had

the opportunity or the blessing of being at a company that's, I know we all generically use the stamp term, but being at a rocket ship company that is literally hockey stick scaling up and to the right like crazy Grove. You got new customers coming on every day, new users. You've got more employees coming on sometimes than you do even on the other side. It's wild, right? It's really, really wild experience. However, for many of us, right? I told you, I shared with you my experience at ACV auctions coming back to your Buffalo being a part of that company, PRPL. Guys,

It's also this incredible masterclass and this incredible education set around.

all sorts of things you should do, all sorts of things you shouldn't do, or all sorts of things that if and when you ever had a time in the future that you were going to do your own thing, build your own company, build your own product, your own service, your own team, gave you a lot, like almost like a laundry list of things that you'd want to kind of bring into that experience. So I love that for you, you know, early on you got to see some pretty incredible things that probably seeded or at least prepared and really equipped you for what you and the team are doing at Assemble today, but super duper cool.

Adrian Brady-Cesana (06:09.191)
Let's jump into the first pillar of team, man. You gave us the high level on Assembled. And guys, for those of you that don't know Assembled, as Ryan just said, they're building something that's pretty incredible, AI for world-class support operations. And we live in a world right now, guys, where we are trying to blend not just the workforce and all the people we have, but we're trying to blend intelligence and all the data that we have. Plus, we're living in this world where we're all being kind of whipped in the butt to try to figure out what the hell our game's going to be or how we're going to begin to master a game.

or how AI sits inside of all this, or us being able to leverage AI inside of all this. And it's not easy. right, first question for you is, jumping in the first pillar of team. What did the early days at Assemble look like? Once you started to figure out you had something on your hands, you started to realize, there's something here, there's the space, there's these needs. Plus, you came from this background of like, you already saw what some of the biggest companies or some of the fastest growing companies went through as it related to being able manage and scale.

you know, their customer support office. What was like the early days for you? Like when you started to even think about how to build a team, what were some of the first couple cornerstones or what were some of the first big bets or investments you knew you needed to pull around you in those early days to figure out how to kind of get assembled from ideation or from what you had gotten into from a product market perspective or from an early success point to where it is today. I'd love for you to just spend some time talking about team and kind of bring us through the phases of as the business has grown, what different types of roles

what type of levels of sophistication, what type of people you needed to pull around you to build this business.

Ryan Wang (07:45.443)
Yeah, think it's, everything comes full circle, right? Because in the early days, it was all about creativity and it was all about wearing multiple hats and it was all about, you know, assembled as my team and title. And what we mean by that is even like, you could be in sales, you could be in customer support, you could be in engineering, you be in product, but early on, the teams were just business and product.

Like there was no distinction between each of these kinds of specifications because we needed people to be doing everything and we needed people to know all parts of the company and product would go talk to customers and business would go understand what was going on in product. And I think it all comes full circle actually today because you know, this creativity is wearing a lot of hats. You have to have that in the AI world. And actually, know, a lot of the way that we build the team even today is

is asking, you know, we've got all these kinds of classic specializations that companies of our size and scale, you know, that they get to you, but, but, but AI is a lot of it up. And, and I'll give you the example where even in support, we're hiring for folks who are, uh, uh, able to write SQL. We're hiring for folks who are super creative. We're hiring for folks who have like dabbled in, in, in vibe coding and AI tools. And so, um, it's actually the case already that as of two weeks ago,

More than half of the support team has contributed production code to our code base. And we've built the systems to be able to do that. And we've looked for people who want to do that. I just think up and down and across every single role, you're going to have to look at that. We even did the study just broadly across customer support, where I posted this article in Fast Company, where it analyzed 21 job listings. And just even within support,

75 % of the postings, they said you need experience with AI tools or automation systems or conversational AI systems. 50 % of them, they said you need to know how to use an API, you know how to do debugging with SQL. And if anything, I thought that was, you know, kind of illustrative of the stretching of the skill set, but even like, again, that core characteristic of initiative and wearing a lot of hats, like that's the big thing that we're actually looking for is can you learn all this stuff on the fly because you may not even need to know about.

Adrian Brady-Cesana (10:02.6)
Yep.

Adrian Brady-Cesana (10:07.059)
Yeah.

Ryan Wang (10:09.856)
APIs or how Python works these days. As long as you can use Claude and figure it out.

Adrian Brady-Cesana (10:12.114)
Yeah.

Isn't that right? Right? It's true. And wait, real quick, Ryan, on this front, I've been calling this out more on recent episodes, when you guys, for our listeners, go check out Ryan's Bits Bits Assembled on LinkedIn. All of us go there every single day. You see the home, the about, you see people, you see jobs. When you guys go to your team today, Ryan, it's interesting, and you go into the people section, there's a couple interesting things that jumped off of me. Number one, what they studied. So from your team, the incredible team that you're building today, Assembled, computer science,

Computational science, economics, business, psychology. Okay, awesome. And then the other thing that jumped out at me was what they do. So engineering, sales, business development, information technology, and ops. And guys, it's really funny because what I'm starting to notice is like what you just said, Right there? You can tell very quickly just by going to LinkedIn what...

modern, poised real well for the future, poised real well for where we're already in, this whole new change, whole new revolution world that we're in, versus traditional companies where they don't necessarily have a smattering of those things. They're typically very, very, very pointed. There's like three or four core areas that everybody that did. So for example, if you're looking at an automotive company, everybody there, it's either automotive sales or like automotive operations. And so what you just said that I think is so spot on is tomorrow's founders, tomorrow's

business leaders tomorrow's guys and gals are going to build the best and the greatest companies on planet earth in the future there's this notion of having an extremely diverse a well-rounded

Adrian Brady-Cesana (11:50.963)
Just spread type of team that over time can literally rinse or wash off on each other. They can learn from each other. You can have the salespeople learning from some of the engineering folks. You can have engineering folks learning from some of the salespeople around how to storyteller, how to get something exciting, or how to make somebody really passionate about some math that maybe doesn't sound very hot and sexy in beginning, but maybe that math is the thing that literally equals the conversion juice that the whole damn business is running on.

So I think that's really cool and I think it's also cool that like you're right, man We're living in a world right now. I know it then for our listeners I'm sure you guys are probably hearing me point this out again and again and again But like this is a world where the rules have changed real quickly having just really bright people or having Really really bright subject matter experts in specific areas kind of going away instead We need to get way more comfortable with this idea of bringing people into a space where they might not ever understand the advanced mathematics of machine learning

But you know what? Maybe from some of the things that the ML team can teach us over time or that can build playbooks for us to be able to story tell upon, that rub off starts to happen. And then maybe the next 100 sales calls you go to, you're not a machine, do not get me wrong, you are not a machine learning scientist, you are not a data informatics expert, but you've been.

you've been groomed or you've been positioned or you've been taught by and so then it does make you sound more right for the job or more right for the task at hand. I love that that's kind of how you guys went to it. What were like, putting you on the hot seat here, what was like the first, what were like the one or two roles that you couldn't fricking wait to be able to hire for once Assemble got to the size and the place and the space and then where you could afford it. What was like the one or two roles that you were just like salivating to get around you in the early days of building Assemble?

you

Ryan Wang (13:43.683)
I think it's always been just people who can write more code. And I want to be kind of particular there because I almost said engineers, but it really is people who can write more code because just as a software business, there's no replacement for being able to ship more product, to be able to do it quickly, to be able to do it with high polish, to be able to do it in a way where it's representative of customer problems and not just kind of made up in a lab somewhere. And that's really hard to find the combination of people who...

Just have the technical ability to make it work. That's the thing about software, it has to work. And be interested enough in the problem to say, I want to go to a contact center in Cebu and follow people around and check it out in order to not just hear from a product manager, read a doc, here's what we have to do. But now I have my own mental model of what it is that we're trying to do.

Adrian Brady-Cesana (14:19.249)
Yeah, yeah, yeah.

Adrian Brady-Cesana (14:40.467)
Thanks for watching me.

Ryan Wang (14:42.806)
That just makes the software a lot better. So I think that was always the challenge. That was always like, how can we produce more software at a faster clip that's higher quality? But again, the magical thing, the amazing thing is that now we live in a world where you don't need that technical ability. It's actually more about how can you express...

this is the problem I'm trying to solve. If it's inside of workforce management, there's a spreadsheet that we're trying to automate the allocation of volume to different BPO's or to different AI agents. How do you bring that to life? You're less constrained by do I know Python? How much code do I write? How sophisticated am I at that? And as long as you can describe it well to an AI agent, and as long as we have kind of enough guardrails around at the company to say, well, you're able to actually take that into production.

That's actually where a lot of our focus has been in terms of pointing the people who are really deep in engineering. So I think it would have been, it is, it always has been, I think it still will be people who can produce great software. The skillset for what is actually required to do that is changing, but it goes back to deep problem understanding.

Adrian Brady-Cesana (15:51.257)
That's it's changing rapidly. It's changing rapidly. You're absolutely right. You're right. There's gonna be a whole new set of playbooks that is brought to the market around how different types of people and different types of brains coming from different types of backgrounds are gonna solve some of those problems. It's super duper cool. Right.

Let's jump to the second pillar of tools. We spent a couple minutes kind of talking about, number one, obviously, let's talk about the assembled tool. But I'd also love to, if you're able to share, I'd love to kind of understand, how did you guys go about thinking about what type of tech stack you needed to build and that you needed to construct and you needed to invest in to be able to build a business? you started to see product market, you started adding new customers, you started seeing new real world daily application of the tools, the goods, the bads, all that. What was kind of your...

team's early strategy or early type of view in terms of what tools did you guys need to build assembled? there specific age-old driving test tools that you guys knew that you used out of the bath? Did you build everything in-house? So we've a couple minutes talking about tools.

Ryan Wang (16:52.44)
Yeah, we had a big debate at the very beginning of the company of what should the programming language be? What should the tech stack be? programming language is one of those classic, you can't really reverse that decision, right? Like you go look at Facebook and they had made a very early choice to use PHP and they had to of build around that. Our debate was around, was it go?

which is a language that was built at Google and really for simplicity and for professionalism. And it's got a really good standard library where you don't have to think about, OK, I've got an HTTP library that I have to download that somebody else built or a debugging library that somebody else built that I have to install to make it work. All the tools around the tool come out of the box.

Or were we going to do Java or kind of on the complete opposite end of the spectrum, we were going to do Haskell because it was kind of like a really cool, fun, nerdy, theoretical thing. We land on Go and we land on Go because it's the kind of underlying principle across all of our tooling decisions have been just keep it simple. Like keep it as simple as possible. Make it easy to learn.

have it be a professional stack. And that's really served us well now, again, everything coming full circle because now that we're writing a lot of code with AI, now that we're doing a lot of vibe coding, we've got a broad swath of the company, half of our supported team writing production code, it's actually, you want simple patterns. You want a language that AI can easily produce. And that's worked out really well for us.

Adrian Brady-Cesana (18:34.715)
Yeah, 100%. Absolutely. That's awesome. Was there, because you guys are unique, You naturally, you showed up in the early days of assembled already with being extremely comfortable in a few different areas. Number one, technology. Number two, mathematics. Number three, machine learning and the early sets of AI. Not even really the early sets of AI, because you were thinking about machine learning as it relates inside of this canopy of AI. Was there?

Was there any part of the customer journey or any part of the growth process that you know, or the growth phases that you saw ahead of you that you knew you were going to have to kind of grab some out of box support and that you already knew, you know what, we're not going to screw around with that. Maybe with the CRM we're going to go right to the best of class. Was there like a couple areas of the tech stack that you knew that you just wanted to basically pull something out of the box that you and the team were comfortable with? And if so, if you could share it, which of those tools were there? Or let me rephrase, if everything that you've been going through now in last eight,

plus years of building assembled plus all of your other earlier career experience. Still like an area that you just would tell founders right now they're listening like don't like just focus on some of the important things in front of you.

borrow for now anyway some of these other technologies that you're not going to build any better. You're not going to build a better CRM. You're not going to build a better ERP. Was there any of those types of technologies that now that you've been in the game this long, you'd kind of tell up and coming founders, just kind of start there. And then if you want to build something natural or you want to build something organic or custom after the fact, go ahead. what would be like a tip or two for some of the technology that you learned? Just go ahead and use some of the existing stuff that's out there.

Ryan Wang (20:13.304)
We've tried to build as little as possible in-house. We've tried to be incredibly uncreative about how to run the business. So for example, CRM, we're on Salesforce. And before that, we were on HubSpot. For marketing outbound, we're still on HubSpot.

Everything you go up and down like how do we communicate in slack? How do we you know use documents its notion? It's incredibly vanilla and The thought has been from the very beginning There was this influential blog post from from from from years ago is like innovation chips You don't want to spend your innovation chips on on stuff. That's kind of like sorted out like you want to spend your innovation chips on How is our AI agent solve, you

Adrian Brady-Cesana (21:01.125)
Absolutely.

Ryan Wang (21:06.508)
more resolutions than any other AI agent. How does our co-pilot make people way more productive than any other co-pilot? How can our workforce management system eke out another two, three, four percent of forecast accuracy of schedule generation? So we go super deep there. We've got a custom stack around how we do RAG. We've got a custom stack around how we do convex optimization, how we do schedule optimization across 20,000 different people. That's where we spend a lot of our time, but for everything else, it's super, super vanilla.

Adrian Brady-Cesana (21:08.743)
Yup. Yup.

Adrian Brady-Cesana (21:20.509)
Sure, sure.

Ryan Wang (21:37.103)
The one that we failed at following that principle was feature flags. So very early on we thought, my gosh, it costs so much money to pay a company to do feature flags. Why would we spend 50K on software for feature flags when it's just like on or off? And we did that and today we suffer from it, right? We've got like 600 feature flags and the missing functionality and we have to

Adrian Brady-Cesana (21:41.651)
Okay.

Adrian Brady-Cesana (21:54.323)
You're right.

Ryan Wang (22:03.47)
use our own internal tool to turn them on and off and it slows us down. It slows us down like probably by five, 10%. And it's like, we should have just used a feature flag tool as well. Something super vanilla.

Adrian Brady-Cesana (22:11.119)
Interesting. Interesting.

Yep. That's an awesome advice though. That's awesome advice. And that's something a lot of people aren't super candid about, which is like, any one of us is we're building our business, whether we're building the next billion dollar thing or whether you're building the next million dollar thing. There's certain things that happen or there's certain tools or certain use cases or certain investments that, man, when you go back and you think about them.

Five years later, you're like, what the hell was I doing with that? I wasted a year or two on trying to think about something that didn't really have the bang for the buck that it could have had. But more importantly, to your point, it's not necessarily chasing the whole thing that you're out there in the world trying to become number one, the best at or have the best solution for or the best possible go-to solution for. So like, that's smart, man. I think we do see, great, you see companies doing this a lot. I mean, I don't know about you, but there's times I'll jump off of a demo and I take a lot of demos. Obviously I take a lot of demos.

because I like to learn. like to learn how others are selling, how others are here, how other people are doing their pitches, their stories, what's good, what's bad, how are they grabbing information, how are they catching my ear, what what was, what like what did I just drop out with 30 minute demo and like I'm thinking about it now for the, and it's funny because like.

More often than not demos this day and age, I do leave a bit confused. And I've been doing this stuff for very long time, brother. Where like, and you too, and I know you know what I'm talking about, you think you're talking to like a company that's there to help you with like podcast scheduling optimization. Pretty fricking straightforward, right? Pretty unsexy, but pretty straightforward. But then you get off the call and you're like, wait a minute, I'm sorry. they can do my schedules, they can do my invoicing, they can do my editing too somehow. I talking to a scheduling program that also somehow knows how to do it. And then you get off the call and you're like, you're almost confused as to what that,

Adrian Brady-Cesana (23:50.453)
that one big thing is that that business is out there in the world trying to disrupt or trying to bring to their client base or trying to bring their users. So you're right. It's something about the power of focus, man. Something about the power of focus early and then something about being able to understand what you should and shouldn't be investing and focused on early on. That's a super, super great piece of advice right there, Right, let's jump into the third pillar of process.

Okay, everyone's talking about AI, AI, AI. And that's awesome. It's great. We're in a beautiful part of the world. It's a little scary and a little, there's a lot, lot unknown, but let's, let's call it what it is. For the people that are, are, are, are facing this thing head on and we're rolling our sleeves up and we're building with it and we're learning about it. We're going to, we're going to be the folks that are going to, are going to be in really, really good shape. Fast forward, you know, a decade down the line, but with process, this is an interesting one.

Normally when I ask this question, it's usually asking the guests, know, Ryan, tell us about how you've sort of thought about processing your career, wrangling everything from living playbooks to SOPs and how you set expectations or maybe even how you built out your KPIs and the things that you measure and manage. In AI land, what I'm learning is process is a little bit different because I think what I'm learning more more about is living playbooks, knowledge base, FAQs, general stuff.

substance around what people need to know, ought to know, should know, or will have to learn, this is a bit of fuel. This is a bit of fuel for AI and for all AI engines, meaning to be able to dump massive amounts of incredible, incredibly select geared information into certain fuel tanks, for lack of better term. That's what ends up being able to drive certain AI solutions in a very specific, meaningful type of way. So for you, and I'll stop there, but with you,

How do you kind of think about process? mean, I know that you probably still have to have all sorts of day-to-day process for running assembled. Your team's getting bigger now. You guys have people all across the world that are working together. You got customers all across the world. How do you sort of think about that third pillar of process? And has there been anything that's really kind of changed as you've built assembled and as you've grown this team and you've grown this business and you're learning every day? How do you sort of think about process now in today's world as we're moving into this AI-focused world?

Ryan Wang (26:09.164)
Yeah, it's really interesting because we have almost a mirror into our company because when we're deploying our AI agents, right? We're going into companies and they're always asking us like, how can we automate more faster? And we go, great. You mentioned knowledge base. What's your knowledge base look like? Where are your SOPs? We can take that and run with it. And the companies that have like a great...

organized, well categorized, living playbook. It's updated and it's fresh and it's not out of date. They're getting really quick gains out of AI. You can snap an agent on top. You can integrate it. You can make it agentic to your tools. You can run it through your SOPs. Agentic workflows is what we call it. And all of a sudden it's off into the races. And then there are a couple others and this is...

And actually it should be, I should say it's more than a couple. This is the most companies go through this where it's, okay, well, we've got knowledge over here. You've got knowledge over there. We can't really tell which one's the source of truth. Can you just go through our tickets and figure it out? Can you just go through our Slack? Can you build the processes for us? Can you tell us, know, what should the operating procedures be? And then your AI agent just answers it and...

Adrian Brady-Cesana (27:30.547)
Get ready.

Ryan Wang (27:34.229)
Usually what we're telling them is we can do the best we possibly can, right? Like the interesting thing about AI that's different from machine learning before is that, yeah, you can just drop a lot of stuff in. That's surprising. But it's still not perfect. It's still not like, hey, give us all of your tickets. We've got 2 billion support tickets in our database. We can't just like magic wand tell you, here's how you should run your business, right? Here's how the support agent is going to answer.

Like you said, the value of documentation is actually higher than ever. So like if I think about process and I think about, you know, what is the outcome of that? The outcome of that is a really clean, really well-structured, you know, a really good way to say, okay, these are the things that need to be done in this business and it's repeatable and it's not living in somebody's head. It's not just like we're making it up every single time from scratch.

It is there on a piece of paper and you can reproduce it. That's so much more important than ever. And so we've learned that from our customers. We've learned that, you know, that can take an AI deployment from like months to two weeks to two days, depending on if you've got all of the, you know, documentation in a good place. And process is the system by which you have that or maintain that.

Adrian Brady-Cesana (28:54.845)
Yep. Yep.

I couldn't agree more with it. know sounds crazy, but we still do a ton of work with our clients here at CXC. Usually on the partner led services side, one of the first things we'll start with is whether it's HubSpot, whether it's Intercom, whether it's Assembled, whatever. Understanding where you just said it. Where's the initial playbook? Hey, where's the initial playbook that some human, like not out of the box, not out of the box builder based content, where's like, when you guys got this tool and you spent like 35,

days implementing it? any humans like write shit down or was that just... or record it? Here's the other thing guys, be smart. Some of our younger listeners, you don't have to write anything down anywhere. Record that shit, leverage the transcriptions because guess what? If you got four smart people sitting in a room walking through HubSpot implementation or if you're walking through how you're going to integrate an awesome tool like Assembled into your HubSpot or into your intercom, make sure you're recording those three or four working sessions

internally with the team because then guess what? After those three or four one hour working sessions.

the amount of substance in terms of the actual sentiment or the actual verbiage that was used, dude, it naturally starts to lay out outlines for what a playbook might look like for that. And then being able to connect these things accordingly or being able to understand which areas or which information can just stand as they are because it might be more black and white stuff. Connect cord to port and then plug port into wall and then it works. Versus some of the more complex stuff where it's like you're trying to get sales

Adrian Brady-Cesana (30:31.615)
people to understand when they hear certain things, here's where your brain needs to move in form of that where that conversation opportunity might lie. Or maybe the operative side, when you start to see certain ticket levels rising or reducing, maybe we start to go a little bit deeper. We think about what does that mean upstream or downstream. And then lastly, just like in the world of all the guys and gals that are going to be building from this stuff, the more information or the more substance or the more general sentiment that we have, theoretically, the tighter we can start to

some of these things we can start to codify or we could start to sync together. I think it's something that's really, really kind of interesting. And it's funny, I think that process question, I think it's gonna change rapidly, man. I was listening to some old episodes the other day. Nobody talks, it's funny.

The amount of change of this one question in just the last 50 episodes has been wild. Because I think you have people that hear that question and some of them they think about it half of their brain is still in traditional like just customer business leader land where they're thinking about, yeah, I had to build my playbook and then I had to understand which things matter, da da da. And then now you fast forward into like, well the process also equals essentially content fuel. Or sorry, content becomes fuel to drive said

Said AI needs and products and Willie's and where's we're building so so supercar don't know just kind of interesting and it makes me think like you know at what point do businesses start to really understand that end of the sub is almost like a full-time Informatics internally and making sure that our tribal knowledge is so well chronicled or detailed or documented or or the history of it is so well That you could almost derive Future answers from it just like any other history example. I find it fascinating, man

Ryan Wang (32:17.922)
Well, I'll tell you, it's not something that you think about when you think, you know, what is a process that we should institute, but we were super maniacal in the early days of every single meeting. You had to take really good notes and now it's, you have to record as many as you can, or if you can't record it, you know, put it in granola or if you can't granola it, you know, type up the notes and share it. But every single meeting that's ever happened at assembled, there's notes sitting somewhere.

Adrian Brady-Cesana (32:45.032)
Yeah, right.

Ryan Wang (32:45.134)
And inclusive of internal meetings too, by the way, somebody gave me flack in a meeting earlier today. It's like, hey, who's taking notes here? was like, oh my gosh, what? But that is the input. That is the fuel for not just, like, you know, in the early days when we thought about this and when we would bug people, we would go, you know, I gotta take notes, so I wear your notes. You we would really get on at each other about it. In the early days, it was the thinking that, okay, this context will be useful.

Adrian Brady-Cesana (32:52.765)
Yeah.

Ryan Wang (33:14.392)
for somebody in the future, we could not have predicted that today that context is the term. Context is exactly the thing. Like every single customer conversation, every single internal meeting, that's producing context that then AI can run through and summarize for somebody later when they want to go ask in a Slack channel, what are the themes of all of our win reasons? If the notes, if the meeting transcript, if something exists, it can run through that. If it doesn't exist, it lives in your head. I've got to go have a meeting about it, right?

Adrian Brady-Cesana (33:19.549)
Yeah, 100%.

Adrian Brady-Cesana (33:31.059)
100%.

Adrian Brady-Cesana (33:44.871)
Yep, absolutely, 100%. And then I just think the other thing too is just like, everybody always uses the term work smarter, not harder. But we haven't even started to scratch the surface yet in terms of ways of almost automatically or asynchronously being able to literally deliver or update things that are important within a business, meaning.

Now don't get me wrong, this is a little bit more complex. So what I just said is not that easy. You gotta think about governance levels, you gotta think about who knows what, you gotta think about what your meetings are, are we even talking about dumping into the pool here, we dumping everybody's, we can't be dumping Ryan's meetings into the pool versus, you know, gotta be thoughtful about that. I'm saying everyone would have different rules around that. But.

The reality is we are there, Ryan. And then the other reality is the best companies that start to understand how to leverage the day-to-day things and activities that are already happening, you're already doing your internal meetings. You're already doing your hundred outbound sales calls per day times whatever your number of salespeople are. You're already doing your inbound number of conversations with existing customers 50 times a day times however many support or success. Guys, the stuff's there. like, especially for our listeners that are maybe not

at the level of assembled or you haven't been eight or nine years into building something, you haven't worked with some of these incredible customers that Ryan's team's working with, these are going to be some of levers that make the difference between either getting there faster or frankly out winning or out beating some of the competition that's around you in your area because a lot of the answers are there, a lot of times the things to build, they're already sitting there, a lot of times the things that you should never build are already sitting there. So I just find that really interesting, man. It's a question I'm going to keep asking.

Right, I'd to jump into the fourth and to the final six pillar of feedback. You know how we break this one up, brother. I'm going to ask you the first part of your answer. I'd love for you to spend a minute or two talking about...

Adrian Brady-Cesana (35:40.871)
What are some of the ways as you've grown Assembled, as you've grown the Assembled team, and as you've grown the Assembled customer portfolio, what are some of the ways that you guys have really, really tried hard to leverage your customer feedback? And then that'll be the first part of question, and then I'm gonna ask the same type of question on the employee side as well. So just spend a couple minutes talking about feedback.

Ryan Wang (36:00.29)
Yeah, on feedback, I think we just had this very early belief that, well, you always want more of it, but that the way to get more feedback from customers was to one, ask them for it, and then to two, do something about it really, really quickly. Because then when you ask for people for feedback, they say, OK, you could do this better, you could do that better. Most of the time, people just, it's like, OK, thanks, and then do nothing with it.

And you can tell when you've given somebody feedback and you've been really thoughtful about it, it's like, okay, you're not going to do anything with it. Like, time you asked for feedback, well, you know, what did you do with my initial feedback, right? So we were really maniacal about, you're to give us some feedback and we're going to just act on it really, really quickly. And we had a bunch of tools for this as well. So we have a public feedback board called CanE. We can't keep up with every single one anymore, but in the early days, it really was every single thing somebody told us.

we would do something about. would either fix it or we would explain, hey, you we thought about it. Here's why we're not going to do it. Or we would say, hey, you know, that's part of a bigger initiative that like two weeks from now, make sure to follow up and tell you, hey, we solved it. So I think all of that led to this positive flywheel of we're asking you for feedback. We're doing something about it. You're seeing that we're doing something about it. And then we'll get more.

Adrian Brady-Cesana (37:23.719)
Yeah, I think nothing, I've said this for years, it's number one, two things. Closing the loop, don't even bother asking for feedback, guys, unless you have the ability, the capacity, the know-how, and the general yearning to answer the damn person, because it really does. You are spending social capital in an indirect way if you're not doing that. I'll keep it simple there. But then the other piece to it too is just like, you're right, man, where it's like...

If you already know which areas in a product or service you're doing well...

then maybe ask more pointed questions because more pointed questions might make an individual not have to ponder around some of the stuff that you just kind of laid out, right? Where it's like, I mean, I don't know, I like you. You can see I've been paying your bill for two years. I talk to my rep every single month because there's at least three things every single month I need to figure out with your rep. So I spend that time, you can see that I've been doing. And then also don't poke on the obvious things. So like if you have one of those situations where some customer literally is every month they're hitting their

CS touchpoints they're doing there. They're doing QBR is fine. If they're every every quarter, they're giving you a great QBR report if once a year they're actually giving you an executive sponsor for maybe some type of like maybe you guys are doing like an executive and a council type of thing where You get really really good at understanding what the whole customer base is thinking about But at the same time right you're right You don't want to waste that stuff like if you don't have anything that you really need to poke a prod for that you're gonna be able to do something with and then feed back to him best example ever would be sort of like the idea like

Unless you guys are starting to build like your Spotify into the year recap and unless you're going to like that level of like hey from all the feedback that we've asked for this year here you go like because that's like that would be an extreme example but like Be thoughtful of when and where you spend it I know I'm one of the few CX people that thinks that like you know surveys man surveys have a place and sure that they had a time and they have they had a time in a place but like So a lot of people want to ask surveys about things that are sitting right side of their user data and I'll say to these clients will be like wait a minute Why do you?

Adrian Brady-Cesana (39:26.069)
Why don't we just go to Mixpanel? All the answers are right there. Well, we want to make sure that they know that we're... Why? So like, sure, just be thoughtful around what types of things you're asking. What about on the employee side? The employee side might be different for you. You work with some really, really smart people and then you also had to get people on your team that wanted to work in the future of work. So like, that's different, man. So like I joke around all the time. Like the people that are going to be some of the best and the brightest and the ripest for like tomorrow's leader.

companies, those people are going be hard to keep around. Those are super smart, valuable, really, really interesting people that are going to be a part of building this ad. So what about on the EX side or on the employee experience side? What type of things have worked well for you?

Ryan Wang (40:09.368)
Yeah, we do this basically every year, team NPS survey, and we run that through lattice. And it's kind of the same principle applies to customer feedback. We don't want to ask people for feedback that we're not going to do anything about. You just never hear from people again and they say, here we're running the survey again and nobody's going to do anything about it. So you can kind of see the responses degrade over time and then over.

Adrian Brady-Cesana (40:26.535)
Yeah.

Ryan Wang (40:39.144)
You ask, okay, how come nobody's giving us any feedback? Well, because like way back when, they gave you some really thoughtful stuff and you did nothing about it. So we run this annual survey. We try to do something about all of the big rocks. so one of the things that we did in the most recent one was we were just way outgrown our New York office. And we just heard so much that was like, it's just way too cramped. People are sitting on top of each other.

We had some short-term solutions where we changed up the in-person requirements and the in-office requirements. And then the really big one was just to find some space in the budget and say, gosh, we've heard so much about this and 30%, 40 % of the companies in New York, we're just going to pull this forward moving offices and go find something by the end of the year. So we did that super, super quickly, found a new office.

Got enough space, like really focused on having enough phone booths. Like that was a big problem. You know, made sure it was easily accessible. That's easier in New York to be close to Subway. But then the last piece here, I think it's like following up on the feedback. And I think that's, that's, you know, an eternal challenge for any given company is internal comms. And, you know, I always shied away from that term because it always felt like that's really corporate. That's really big company. Why, why, why should you?

Adrian Brady-Cesana (41:34.877)
Sweet,

Adrian Brady-Cesana (41:40.275)
Yeah, right.

Ryan Wang (41:59.023)
focus on internal comms session, you need an internal comms person. But what really becomes true is when everybody's super busy, when you're in Slack and you're getting ding, ding, ding, ding, ding, got like notifications going on and like stuff, stuff's always happening. You know, people don't really notice, you know, people don't naturally notice. It's like, Hey, you know, we, we, made all these changes, you know, uh, we fixed these, the system, the set this, you know, field and Salesforce that was slowing you down. If you don't then specifically go say,

Adrian Brady-Cesana (42:06.419)
100%.

Adrian Brady-Cesana (42:13.127)
Big doodles, yeah.

Ryan Wang (42:27.32)
Hey Bryce, thanks for the feedback on the next steps field in Salesforce. We fixed it. There's a 20 % chance he notices and goes, wow, that's fixed. know, people are listening to my feedback, but more likely than not, he won't, right? And you know, at the biggest level and at the lowest level, you know, in terms of like the size of these things, like you gotta go tie things together, right? Tell people, hey, thank you for your feedback. Here's what you said, here's what we did.

Adrian Brady-Cesana (42:30.707)
Right. Yeah, right, right.

Adrian Brady-Cesana (42:38.547)
You

Yeah, right.

Adrian Brady-Cesana (42:52.21)
HIDAL together.

Ryan Wang (42:57.454)
You know, what else, right? So I think that's the step that we've always skipped and that, you know, for customers and for internally, for assemblers, is probably the most important one. It's like, you do something about it, go make sure you tell people.

Adrian Brady-Cesana (43:12.965)
Amen. I think that's spot on, brother. Ryan, this has been absolutely fantastic, brother. Before I let you go, a couple things. Number one, anything big that you want to shout out, anything big that you and the Assemble team have coming up, whether it's events or new product features or anything like that, anything that you want to shout out to the CX Nation to make sure people know about it, that's number one. And then number two, where can people find out more about Assemble or try to get in touch with you or someone on your team if they want to learn more about this incredible tool?

Ryan Wang (43:40.015)
Yeah, we just did a really big launch around AI-powered schedule generation. So if anybody that's had to create schedules and figure out how to juggle the labor laws in this country and the preferences of these handful of folks and folks across different offices and so on, across different skill sets, now there is a way for, and you don't have to know, 20, 30 years of a workforce management tool.

We have AI in Assistant where it'll just help you to tell us what you want and run you through it. And you can generate schedules for up to 20,000 people at once. That's our biggest deployment. We've got a really cool thing coming up on our AI voice agent. We now have the ability to support multiple brands. So many of our customers, they're actually running different types of sub brands within one broader company. And so you want to have different tone. You want to have a different type of style.

You have different types of workflows that you want to run through, and you want to be able to run that all out of one single brain. So now we have the ability to support multiple brands within a single studio. Where can you find out more? You can find us on assemble.com. It's just a super easy way to book a demo. You can find us on LinkedIn. So Assemble the page or myself, we're often posting about the future of work. And certainly within customer support, I think we're at the forefront of what's going to happen with human and AI agents.

So give us a follow and let us know what you think and what you're seeing. We always want to hear from you.

Adrian Brady-Cesana (45:09.157)
I love it. Wang, it's been an absolute pleasure having you on the show, my friend. I can't wait to see what you and the team at Assemble do next. And we're going to be following you, man. This is a super cool company you're building. You're working with some incredible customers. got some incredible people on the team. Super, super cool, man. You should be proud of yourself. And I look forward to our next conversation in the very near future. Thanks, Ryan.

Ryan Wang (45:25.634)
Yeah, likewise.