Latest / State of the AI Union / Punjun Bhatnagar & Jeff Gibson: Filling AI Cracks with Gold - Kintsugi
Transcript
- 0:03All right, everybody. Welcome back to the State of the
- 0:05AI Union, and I am your host, Laura.
- 0:08This episode is part of our new startup series where we sit down
- 0:11with people who are building the next generation of AI native
- 0:15companies. And these aren't just startups
- 0:17slapping AI onto old software. These are founders who started
- 0:21from a blank state and asked like, if AI is the center of
- 0:24everything, how would you build a company differently?
- 0:26Today, I'm so happy to be joined with by a special duo from King
- 0:31City. We've got Punjan Bhatnagar, who
- 0:34is the CEO and also Jeff Gibson, Co founder and CTO joined with
- 0:38me here today. And I'm going to give them a
- 0:41second to introduce themselves directly and maybe tell us also
- 0:45a little bit about Kinsugi. So, Pujan, why don't you start?
- 0:48Yeah, for sure. So hey everyone, name's Pujan,
- 0:52one of the Co founders and you have my other half Jeff here.
- 0:56Maybe taxation is in my blood as I like to joke about.
- 1:00My father and my grandfather worked in the accounting and CPA
- 1:04space for combined of like over 70 years.
- 1:08My grandfather was born in 1928. This was pre independence of
- 1:14like India and really like growing up.
- 1:19He pioneered about how when India was finding its like
- 1:22footing, thinking about how consumption tax should be part
- 1:26of so paid a contribution to the initial draft of what is now
- 1:32India's CST and GST bill and how things are thought of.
- 1:35A little bit of a background about me.
- 1:38I'm I'm a data nerd. I love math and computer
- 1:41science. So some stereotypes are indeed
- 1:43true. Grew up in New Delhi, India,
- 1:46part of my life lived in China and Dubai.
- 1:50I came to the US for my undergrad, finished all my major
- 1:55requirements probably in about a year.
- 1:58And I had known that I wanted to be in Silicon Valley and had
- 2:02gone into Stanford graduate program media science,
- 2:06specifically in AI. And then I think during my time
- 2:12at Stanford, one of the things that I really resonated with me
- 2:16was data is king. And as as I was working with
- 2:20applied Statistics and AI and I was working on data sets like
- 2:24squad data set, which was the Stanford question answering data
- 2:27set that later on became the foundation of ChatGPT and large
- 2:31language models. I realized that how the dense
- 2:37state-of-the-art algorithms were making things revolutionary
- 2:42because what we call as large language models, we were working
- 2:47on them for nearly a decade before they got commercialized.
- 2:51Then I got an opportunity to really join Facebook and kick
- 2:56start the machine learning effort, especially for the
- 2:58Stories organization. So became the first machine
- 3:01learning engineer there and launched Stories ranking and for
- 3:08platforms like Facebook, Instagram, WhatsApp, Messenger,
- 3:12and really learned how to build on scale.
- 3:15So by the time I left in 2021 to pursue Kinsui as an idea, we
- 3:20were doing, I believe, 25 billion banking requests a day
- 3:24because people, people open all these platforms so many times.
- 3:29So again, leveraging data to find Nuggets of information.
- 3:34At the same time, I got connected to Jeff, who asked the
- 3:40code to cache architecture expert and he'll introduce
- 3:43himself in just a second. I was an immigrant so decided to
- 3:48take a step back from my job and became a scholar plus a guest
- 3:53lecturer and a student at MIT. Again, was teaching data
- 4:00decision making and a couple of statistical things and when the
- 4:04company got traction, moved from MIT to Harvard Business School
- 4:08and then later on ended up dropping out to pursue this
- 4:11company full time. So that's a little bit about me.
- 4:15I like to say that Kinsugi is not a sales tax company.
- 4:19It is a data company. And the first thing that we are
- 4:22solving for is sales tax, but there's a much bigger vision.
- 4:25On to you, Jeff. Thanks vision.
- 4:28Yeah, so I'm, I'm Jeff, I'm the CTO and Co founder.
- 4:31And yeah, my background is, you know, software engineering 20
- 4:35plus years. But about 13 years ago I got
- 4:40introduced to something called quote to cash, which is really
- 4:43just like the academic term for like the flow of money in a
- 4:46business, right? So like, you know, lead
- 4:48opportunity, quote order provisioning, billing, Rev rec,
- 4:52subscription payments tax. And I worked on some pretty
- 4:55large free IPO projects back in my consulting days, building
- 4:59custom, implementing existing software, fixing processes.
- 5:03And you know, there's a couple of common themes problems
- 5:07companies had and one that became bigger each time I looked
- 5:11was tax, tax payments, billing, revenue recognition.
- 5:14There's a lot of problems. So I worked on the pre IPO
- 5:17projects for Atlassian and then I joined them as their first
- 5:20quote to cash architect in like 2015, 2016.
- 5:24The last scene being the first product like growth company and
- 5:26the largest product like growth company.
- 5:28You can actually buy any software.
- 5:29There wasn't like a billing system that you could buy.
- 5:31It wasn't, you know, there wasn't systems that were
- 5:33designed for even that scale of data at times we would have to
- 5:36like get very creative even how we would we would scale our CRMS
- 5:40just because they weren't really designed to have all those free
- 5:42customers as accounts. And so that was like a common
- 5:45problem. And then, you know, over the, my
- 5:47time in Atlassian, I ran data platform teams, machine learning
- 5:51teams, software engineering teams.
- 5:52I worked on strategy for acquisitions and new products.
- 5:56And everywhere I went in the business, I just kind of
- 5:58observed there was like certain things that made it really
- 5:59challenging. Billing was always 1, tax and
- 6:02payments were always 1, acquisitions and the
- 6:05complexities that come along that are always there.
- 6:07And so my my first start up was sort of building like a billing
- 6:10system, which was kind of Atlassian's product like growth
- 6:12in a box. And through that I observed that
- 6:15tax was still a problem. And I was kind of shocked.
- 6:17I'm like, this was like painful a while ago I expected to get
- 6:20better. And when I looked around, it was
- 6:22just the same sets of players and the same people and the same
- 6:24problems and nobody really doing anything interesting, not a lot
- 6:27of innovation. And so when I just saw the
- 6:29opportunity and again from my background, like, you know, I in
- 6:33Atlassian and others, I was always solving data problems and
- 6:36then they had business applications on top.
- 6:38And so a lot of the things that I was doing at Atlassian, it's
- 6:41just a ridiculous scale. Those same patterns and and
- 6:44approaches actually work really well for things like tax and and
- 6:47compliance. So yeah, when Pooja and I teamed
- 6:50up, it was certainly not my my first rodeo kind of building
- 6:53platforms like this and ingesting and processing these
- 6:56volumes of data. And so I was like, I felt like
- 6:58sort of my career was like leading up to having all the
- 7:01skills and knowledge to build the right thing in the right
- 7:03moment at Kisugi. And so it's been kind of a fun,
- 7:06exciting time. So I'm glad to talk more about
- 7:09it. And, you know, as you know,
- 7:09we're, you know, we are an AI first company and I think you've
- 7:13got, you have some of your questions, you know, you have
- 7:14about like, how do you talk to your customers about AI and
- 7:17things like that? I'm really excited to talk about
- 7:18it because we've learned so much in the past couple years.
- 7:21Yeah. And and you know, it's it's
- 7:23really interesting how you how you said that, you know, all of
- 7:26your experience basically landed you up here in this moment to
- 7:29start Kinsugi. And I think Kinsugi is the art
- 7:31of lacquering, right? Or and filling with gold.
- 7:33Is that correct? Like repairing some gold?
- 7:36Yeah. Awesome.
- 7:37So I'm I'm a nerd when it comes to languages.
- 7:42I, I love learning them. I have 9.
- 7:47But one of the things that like the analogy of like, consider
- 7:52use the art of like joining pieces and like with gold is
- 7:56what we see with sales tasks, right?
- 7:58At the end of the day, you need 360 view of revenue and 360 view
- 8:03of HR to be able to calculate and figure out how much sales
- 8:07tax do you owe and where do you owe it to.
- 8:10And so all of these different pieces of information, they live
- 8:13in different Erps, different billing solutions, different HR
- 8:16solutions that companies implement and they don't talk to
- 8:19each other. And so one of the key insights
- 8:24that we had especially given just like PLG background and
- 8:28experience was the process of implementing this and solving
- 8:32this is, is so cumbersome and so manual.
- 8:35And so it starts being a non revenue generating company.
- 8:38People often times they are like, you know what, I'm going
- 8:41to kick the can down the road and before you know it, they're
- 8:44raising the Series A or Series B or in case of like brands
- 8:48looking to get it hired or raise more money.
- 8:50And the number one question that any any serious VC or growth
- 8:56equity firm brings up your due diligence says are you
- 8:59compliant? And when that happens,
- 9:01especially if like the founders are looking for a payout, it
- 9:05becomes the number one source of decreasing your valuation.
- 9:09So suddenly something that was non revenue generating and it
- 9:13was something that was ignored because the solutions don't
- 9:16really solve for it because they are so cumbersome to implement
- 9:21now comes in bites you when you're at least expecting it.
- 9:24So the unique insight that we had is, hey, we are seeing all
- 9:28these transformative changes in the AI and automation.
- 9:31How do we leverage the changes in the field and Jeff's
- 9:35expertise and saying that we will do all the heavy lifting
- 9:39and create a solution offering which says connect your data in
- 9:43seven simple clicks and give us your permissions in 3 minutes.
- 9:47And we will do all the heavy lifting to tell you where, how
- 9:51much and how and really what you owe to different jurisdictions
- 9:58around the world. And from there on out, like one
- 10:01of the key things that we agreed on on Day 0 of Konsugi was we
- 10:07want to create maximum value for consumers even before we ask for
- 10:11a credit card information just like how Atlassian did.
- 10:14So we do all the heavy lifting. We do all the bulk of the work
- 10:20through our no code integrations.
- 10:22And then we tell you without asking for any money, and this
- 10:27isn't how much you owe. And then you can either use our
- 10:30APIs to put your sales tax on autopilot, or you can do it
- 10:34manually by yourself. And all of that was made
- 10:37possible because of the advances in API technology, the advances
- 10:42in AI and how we process like something billions of rows of
- 10:46data on an asynchronous basis. So, right, that's why it's
- 10:50called Kintuki, and that's our ethos behind the company.
- 10:54Love it. And since you're talking about
- 10:56building the company and leveraging all of just
- 11:01experience, maybe a question for Jeff here.
- 11:03Having having built engineering teams right and systems in in
- 11:09the past, what was different about building an AI native or
- 11:13an AI first technology team? Yeah, I mean, there's a lot of
- 11:17things and some things I, I kind of expected because I, I managed
- 11:20data scientists, machine learning engineers, etcetera in
- 11:23the past. But I think that was always like
- 11:24a part of what you were doing. I think to say like actually
- 11:27that's the whole foundation of what we're doing.
- 11:29It makes it very different who who you want to hire, the types
- 11:32of people you want to hire, the types of, you know, the types of
- 11:35problems you're solving and also the problem solvers.
- 11:38And so I think #1 was just looking for people who had that
- 11:41interest in learning and trying new things and taking a little
- 11:44bit different approach. You know, like it's, it's
- 11:48interesting, like over the years, you know, data and
- 11:51machine learning and AI has changed so much.
- 11:53So much of your time used to be sort of preparing the data, used
- 11:57to be a take like 8-6 to 8 data engineers to get the data in a
- 12:01good state so that one data scientist could get some useful
- 12:03insights. And I think something
- 12:05interesting is looking at today, that's just not the case.
- 12:08Actually, it's way easier to get access to data.
- 12:10It's way easier to get access to compute, to storage, to scale
- 12:13and be all these great tools out-of-the-box.
- 12:16So I think the ratio of how we were thinking about like, do we
- 12:18need software engineers? Do we need data scientists?
- 12:20Like what types of work will be be doing to produce something
- 12:23valuable? I mean, at the end of the day,
- 12:25we're trying to solve a business problem.
- 12:26And so I think one learning was just thinking about what types
- 12:30of people we would need today in this world versus 10 years ago
- 12:33as hiring teams of data genes. Yeah.
- 12:35And, and, you know, I think that here in the series, we're
- 12:41talking to AI native founders that have that have decided, you
- 12:46know, an AI centric company is the way to go.
- 12:49And there are also companies that are trying to build the
- 12:55solution on, on their own, right.
- 12:56I'm sure you've spoken to a number of different
- 12:59organizations of customers that were trying to solve it on their
- 13:01own. They felt like they could maybe
- 13:03deploy an AI agent on their own to solve this problem.
- 13:07I wonder what are some of the what are some of the things that
- 13:13customers like are have to consider when they're thinking
- 13:17about building their own using existing system versus buying a
- 13:22solution like Katsuki? Yeah, Maybe I can take that one.
- 13:25So I think it's one interesting thing is just how fast things
- 13:27are advancing. I think if three years ago you
- 13:29wanted to deploy an agent, I think you were doing a lot of
- 13:32work, you were doing with a lot of things manually.
- 13:35You didn't have a lot of options.
- 13:36There wasn't a lot of great frameworks.
- 13:37But even a year later it was completely different.
- 13:40So I think one thing is you also like you have to think about how
- 13:43close is the technology to what you need and how fast it's
- 13:46changing. If you're gonna make a six month
- 13:48investment in something, it could be in five months you
- 13:51could use the technology to get to do one month's worth of work.
- 13:54So I think something is also considered just where like how
- 13:56mature things are and whether you can leverage them and which
- 13:59pieces. Like today, if you were building
- 14:01a chat box from scratch with the UI from scratch, everything from
- 14:05scratch, that would just be like a tremendous waste of time
- 14:07versus, you know, 3-4 years ago, maybe that would have been
- 14:10something innovative your competitors didn't have.
- 14:12Yeah. And and Jeff, you, you teach
- 14:16pricing at Stanford, right? Yeah, through for a couple
- 14:21different avenues. 1 is for for Starducks, which is an
- 14:24accelerator that can sue you as a part of a couple of years ago.
- 14:27I teach the cohorts about pricing and packaging.
- 14:29It's always it's, it's funny, right?
- 14:30Like people from strong technical backgrounds rely on
- 14:33the things that they're strongest to focus on, right?
- 14:35And a lot of times that's code, technology, data.
- 14:38And it's like it makes it a little bit more scary to
- 14:40approach things like pricing and packaging, which, you know, are
- 14:42just there's sometimes, yeah. And and how has, how has pricing
- 14:47changed or for you when, when you're talking to customers
- 14:50maybe about the value of AI, you know, as they're also thinking
- 14:53about building versus buying? Yeah, it it's changed a lot.
- 14:56I think it's like if you look at traditionally how most people
- 15:00think about pricing at a basic level is like what are the
- 15:02things that cost us money? And then what are the things
- 15:04that provide value? And hopefully those are the same
- 15:06things. And I think with a lot of the
- 15:08things that have happened in AI, these features and the things
- 15:12that now cost you money there may not be directly related to
- 15:15the things your customers find value.
- 15:17Like somebody could perform a very expensive search and not
- 15:20get the result that costs you money, but they may not find
- 15:22value. And so I think the new trend
- 15:24we're seeing in pricing and packaging and a lot of the AI
- 15:27models is trying to actually trying to actually figure out
- 15:30the things that people value when they are, when they
- 15:32successfully complete something, right?
- 15:34When they complete a task, when they complete an objective.
- 15:36And I think this is where people are trying to go.
- 15:38I do think it is challenging because I think that like many
- 15:41features and capabilities are still very early or they're
- 15:44maybe not even the core part of an offering.
- 15:46So we're also seeing combination of hybrid models with, you know,
- 15:49platform fees, consumption, still seeing services like
- 15:54onboarding being built into things.
- 15:56But there's also like now minimum commitments is being
- 15:59used as a model to where it's like, Hey, I want you to commit
- 16:02to a minimum amount of spend, but then you'll consume through
- 16:05usage. And so you'll be able to track
- 16:07what you're getting. But at the end of the day, my
- 16:08sales team needs to be commissioned somehow.
- 16:11So like we need to sort of agree to a number that we're going to
- 16:13get this year. And so I think also just like
- 16:16the the rise of like metered billing and consumption based
- 16:18billing is actually making it very complicated for people to
- 16:22have sort of like hybrid models and adjust their sales
- 16:25approaches. And so I think at the end of the
- 16:26day, people are struggling to communicate to that consumers
- 16:29and there's like 10 or 20 start-ups trying to work on this
- 16:32like billing for agents, consumption based billing.
- 16:36And so it's a lot has changed even in just the past four
- 16:38years. And Laura, like one of the
- 16:42perspectives that I would add to it is like just putting my
- 16:46perhaps like the experience of what I learned from the Cpas and
- 16:50the accountants, my father's background and my grandfather's
- 16:53background. Do you have one, one spectrum of
- 16:56people that like are like technical and builders who are
- 17:01like thinking about how do we do metered billing, how do we do
- 17:04API consumption, etcetera. On the other hand, given that we
- 17:08play in the compliance space, a lot of like Cpas, they think of
- 17:12it as how do we capture the hours of work, right?
- 17:18Because they are all used to like $300.00 an hour, level
- 17:23hours, the level hours. But like with the advent of
- 17:27software, one of the unique things that we see is if the
- 17:30software is working the way it should, it should not take
- 17:33hours, it should not take minutes, it should take
- 17:36milliseconds, right? So how do you capture, how do
- 17:39you move away from this concept of billable hours and move to
- 17:43the concept of how much value are you providing to the
- 17:46customer? And then the other sort of the
- 17:49access that unlocks for a company like ours is when it
- 17:54comes to PLG is some customers may be very different than
- 17:58others. So customer cohort becomes very
- 18:00important. For instance, there might be a
- 18:03segment of users who are free users who are not taking any
- 18:07actions, but they want to use your platform for awareness and
- 18:11keeping on top of things. Then there might be some
- 18:13customers who are price sensitive and they are thinking
- 18:16about, hey, I don't want to commit to a fixed fee, I want a
- 18:20per action model. And then there are some more
- 18:23advanced users or people who are more experienced in the field.
- 18:28These are people who have done like couple of startups or they
- 18:31are like CF OS or controllers of either on a fractional basis or
- 18:35full time basis. And they have seen how pesky
- 18:37sales tax is. And they're all about no one
- 18:41price. I do not care.
- 18:43Make this problem go away. So how do you capture that
- 18:45nuance? And how do you capture that in,
- 18:48in the pricing of peace, especially with AI where like
- 18:51you can now do a majority of heavy lifting in a couple of
- 18:54seconds, if not milliseconds, How do you do that?
- 18:57I think that that is very interesting and that is
- 19:00something that we have fun on a journey to learn as well.
- 19:03Yeah. And so I'm hearing like a lot of
- 19:04flexibility also. Yeah, a lot of flexibility in in
- 19:09giving your customers choice in how the AI is actually going to
- 19:12make the most sense for them. Punchin, did you drop out of
- 19:16both Harvard and Stanford? Is that what I heard?
- 19:19Harvard, Stanford and MIT might as.
- 19:21Well, Harvard, Stanford, everything you you might be the
- 19:25only person that dropped out of all three of those schools.
- 19:28Let's just say I'm a constant disappointment to my parents.
- 19:33So, you know, I, I know you dropped out of those three
- 19:35schools and that's like pretty amazing.
- 19:37And I, I am, I'm just wondering about what you think is the
- 19:44thing, you know, the thing that you're chasing when you're
- 19:48building Kinsugi. What do you think you're trying
- 19:51to fill as far as the cracks here?
- 19:54I think the the problem is that we want to empower local
- 19:58businesses, right? And like Jeff and I, we
- 20:02identified a gap in the market where the the incumbents and the
- 20:06legacy players, they were still API one point O and they were
- 20:11still relying on traditional on boarding methodologies that
- 20:15really left. That that were not consumer
- 20:17centric, that really like left a lot of question marks for the
- 20:20consumer asked you OK, so how much how much money do I
- 20:23actually owe and where do I owe it and how will this all work?
- 20:27So yes, I did drop out of all of the schools, but like you asked
- 20:31an interesting question about what I'm chasing.
- 20:33I think what I want to do is I think of the world as a system
- 20:39of pipelines, a system of pipes, and our world financial system
- 20:43is basically a system of pipes where money flows like water
- 20:47from one place to the other. I want to build the system of
- 20:51pipes for compliance with Jeff at Kitsuki.
- 20:55And one of the reasonings that I had was, look, some impeccable
- 20:59founders have gone to these amazing institutes.
- 21:03I will have my own journey just like they they did.
- 21:07But one of the things that my mom always told me growing up is
- 21:11you have the same number of heads as others, you eat the
- 21:14same food as others and you breathe the same air as others.
- 21:18So what I am trying to accomplish is when you identify
- 21:22the problem, you need to be confident enough that you feel
- 21:26like, hey, I have the skill set of putting a key together to
- 21:30solve it. You also need to be humble
- 21:32enough to realize that if you don't solve it, somebody else
- 21:35will. So you have captured lightning
- 21:37in a bottle and the only thing that matters is you run.
- 21:41And really, like when you're working on a startup, yes,
- 21:44there's a level of risk involved, but these institutes
- 21:47like the Stanford MID and Harvard are not going anywhere,
- 21:50right? So why don't you go and run?
- 21:53And the third thing that I'm achieved and I'm trying to
- 21:56achieve is if, if you really think about the status quo of
- 21:59compliance, there's a lot of human beings in the loop.
- 22:02There's a lot of like just paper pushing.
- 22:05It's a lot of cumbersome things that are needed.
- 22:08And what I want to do is my mom is a, she's a retired economics
- 22:13professor. And in the field of economics,
- 22:15we call this an inelastic problem because there are human
- 22:18beings in the loop. I want to make it elastic so
- 22:23that when I have kids, I can say, hey, this pesky problem of
- 22:28compliance. Jeff and Ivy made this into an
- 22:31elastic problem, and I think that's a goal worth pursuing.
- 22:34I love it. I love it.
- 22:36Now Punjan, just to follow up on that, you know, you were also
- 22:40like the first MLM engineer at at Facebook.
- 22:43What do you think is the future of AI in the enterprises, let's
- 22:49say in five years? That's a question that I
- 22:53increasingly think about. I think right now where we are
- 22:57in terms of like AI, if I were to put my nerdy hat on, and
- 23:01again, I never finished or got my PhD, so I'm not a doctor, I'm
- 23:06going to pretend like that. But AI right now is sort of
- 23:12starting to show cracks on one trying to replicate human
- 23:16behavior in terms of like with the way humans talk and the way
- 23:19the humans interact. So I think you're seeing those
- 23:22limitations in the current state-of-the-art AI models.
- 23:28But what we are also realizing at the same time is the back end
- 23:32task and the automation tasks AI is really good at.
- 23:35So recognizing that, not great at mimicking human behavior, but
- 23:41really good at doing the boring tasks, that's where we are right
- 23:44now. And the reason behind this is
- 23:46like AI has not fundamentally cracked what is called symbolic
- 23:50manipulation, which is why AI is not able to sort of reason and
- 23:54come up and solve mathematical challenges or proofs that we are
- 23:58still waiting to answer, like the current state-of-the-art LLM
- 24:01models. What they do is they predict the
- 24:04likely, the most likely word following based on the context
- 24:09it's given and based on the corpus of training that it has
- 24:12done, which is the entire Internet now.
- 24:14But symbolic manipulation, AKA reasoning has not yet been
- 24:18cracked. Now you ask the question about
- 24:20where I see the field going in five years, in the next 5 to 7
- 24:25years, I believe and I'm hopeful that we will crack that piece on
- 24:30a research side, which would mean that in the next 15 to 20
- 24:34years, that's symbolic manipulation.
- 24:37Once it's solved, would be available in everyone's hands,
- 24:40and then it would have critical down effects.
- 24:42Amazing. So Jeff, you talked about how
- 24:46the future of work looks like for new AI teams as you're
- 24:52building these teams and how work can shrink really fast.
- 24:56I'm wondering what you think about buyers in the next 5
- 24:58years. Like how do you think buyers
- 25:00will evolve as they start to evaluate new AI, AI tools?
- 25:03What do you think should be sort of like top of their mind
- 25:06starting with now? And then what do you think it
- 25:08will be in like 5 years? Yeah.
- 25:10And I think I think it's actually going to return back to
- 25:13where it was even just a few years ago where where people
- 25:16should really be evaluating the business problems they want to
- 25:19solve. What is the cost?
- 25:20What is the value? Is it a fair value?
- 25:22Do I get, I think, you know, when a lot of this stuff started
- 25:26blowing up a couple of years ago, people rushed to put AI
- 25:30features into their products so they could say they had AI
- 25:32features. And that certainly excited
- 25:34investors. It didn't necessarily excite
- 25:37users, right? If you ask, you know, maybe
- 25:38users like how they're like Asian Force rollout is going,
- 25:41it's Salesforce, they may not be as excited as like the investors
- 25:45when they announced Asian Force. And I think that's like that's a
- 25:47reality. And I think that now what we're
- 25:49seeing is actually practical, useful uses in our daily lives
- 25:53and different pieces of software buying, using and evaluating.
- 25:56And it's actually people are focusing on the valuable thing.
- 25:59Sometimes the valuable things don't require any advancements.
- 26:02It just required more thought. And then, you know, user
- 26:05experiment, experience, experimentation.
- 26:07But now it's a lot easier to to iterate and do things that were
- 26:10more difficult for a user, more difficult to build, required
- 26:13more infrastructure. This is sort of like an extra
- 26:16tool people can have to improve the value of the products they
- 26:18already have. And so I think now if I'm a
- 26:21buyer and I'm looking at software, I'm trying to cut away
- 26:24from the marketing fluff and say like, when I use this as it
- 26:27actually solved my problems, does it actually increase my
- 26:29speed? Is the trade-offs of relying on
- 26:33AI features still gonna give me the reliability of the features,
- 26:38right? Sometimes.
- 26:39There used to be features that would work very consistently
- 26:41every time, and maybe you didn't like them, but maybe the
- 26:43consistency is the thing you value.
- 26:45Maybe now they're faster. And I'm also going to think a
- 26:48lot about like what sort of data I'm putting into the system and
- 26:51can I get value out of this data?
- 26:53The systems that are best leveraging AI are the systems
- 26:55that hold the most data. And I've thought about how they
- 26:57would structure them. And we thought very thought
- 26:59about how they can access them, how they can reference them, how
- 27:02they can create them. So this is where I think you see
- 27:04systems like, you know, like CRMS and things like that, where
- 27:08they have all of the data or data warehouses or data
- 27:11platforms, they're able to do a lot of things, right.
- 27:14So I think I'm also trying to evaluate like the maturity of a
- 27:16system, how much data it can handle and what can it do with
- 27:19that data. But most importantly, is it
- 27:21actually doing the thing that I really want?
- 27:22You know, there's a lot of, there's a lot of software where
- 27:24I do not get that excited when I see an AI feature.
- 27:27Yeah, you know, the system, the systems that manage our employee
- 27:30stock options, I really don't want any features in that like
- 27:33I'm perfectly happy if that's just a Ledger and a great
- 27:36experience. So I think there's people will
- 27:38return back to looking at do I get the ROI?
- 27:39For what I'm buying, you know, I think I think you're right
- 27:41because like right now, like customers, customers and I think
- 27:46vendors are rushing to like just here's the here's the AI
- 27:50feature. And we might be just forgetting,
- 27:52hey, what was the thing that we were trying to solve in the
- 27:54first place? So that's great insight.
- 27:56OK, listen, I have a couple of quick fire questions for us to
- 28:01close out. Are you guys ready?
- 28:02OK. All right, Punjan, what was
- 28:05harder to drop out of Harvard stand for MIT?
- 28:08I would say yeah. And the reason straight straight
- 28:14up being I saw the opportunity and when you see the opportunity
- 28:18you you grab them by both. That makes sense.
- 28:21OK, great. OK Jeff, for you AI subscription
- 28:25or pay per use, what would you rather choose for your daily
- 28:29coffee? For my daily coffee
- 28:32subscription. Paper use.
- 28:34I love this question. Pay per use because I want
- 28:39optionality of where I get my coffee.
- 28:41OK, good. All right.
- 28:43OK. So if Kinsugi is the art of
- 28:47filling cracks with gold, what's 1 crack in your daily life that
- 28:52you're hoping somebody will fill for you with gold?
- 28:56Wow. In my daily life, I think.
- 29:01I think if I could figure out a way to get an automated summary
- 29:06of everything that I need to accomplish on a daily basis as I
- 29:10start my day, hopefully in a stacked grand order.
- 29:15So I think that would be amazing as ACEO.
- 29:18What is important is not like they're always 99 things that
- 29:22you can you can do, right? Yeah.
- 29:24But still bring it down to the top three that are needed that
- 29:27require most of my attention. That would be blissful.
- 29:30So like your daily, your daily prioritization agent?
- 29:34Yep. OK, awesome.
- 29:35Jeff, would you rather debunk code for 12 hours or sit through
- 29:39a 12 hour pricing committee meeting?
- 29:43Well, depends on if I wrote the code or someone else wrote the
- 29:46code. If someone else wrote the code
- 29:51then probably not. If it's me then it's my fault.
- 29:53I should fix it. That.
- 29:56Makes sense, Punjan. What's one thing people
- 29:59constantly undervalue in startups?
- 30:03And that could be either talent or technology.
- 30:05I think what people undervalue is getting shit done.
- 30:09As a startup founder, you need to be very clear about who is
- 30:14building, which is Jeff and who's selling, which is Pujin.
- 30:17And nothing else matters, right? There's the builder and he we
- 30:22are both Co founders and we are both equals.
- 30:24But at the same time he makes the final call on what is to be
- 30:28built. So design, product, engineering,
- 30:31R&D goes up to him. I make the final call as to
- 30:35selling. So which means like sales,
- 30:37marketing, customer success etcetera.
- 30:39And I think people often times they flip flop and they're
- 30:43married to the title, but because they want to get buying
- 30:47from everyone, nobody's held accountable.
- 30:49And unless you're able to like just distill it to hey, what
- 30:54belongs in your court versus what belongs in your Co founder
- 30:58court, I think that's fair. Like most of the startups die
- 31:02because they don't have that tier sense.
- 31:04Yeah, So I'm hearing like trust, trust and accountability and
- 31:09moving, moving real fast. Yeah, don't let perfection be
- 31:14the enemy of good. Yeah, awesome.
- 31:17OK Jess, when price is wrong or value is misunderstood, which
- 31:23one annoys you more? So my my biggest complaint how
- 31:29most software companies sell things is they want me to pay
- 31:321st and then do work and then see value.
- 31:35Like it seems so obvious, but if somebody, whether it's a sales
- 31:39or marketing or product, you should be able to evaluate
- 31:42something, be able to see the value and then pay for it,
- 31:46right? And I think that this is
- 31:47partially just because of how contracts work and sales life
- 31:51cycles work and things like this.
- 31:52But I oftentimes I'm like, hey, can I give you a little bit of
- 31:54money this year and then more money next year and then like a
- 31:56bunch of money in the third year if you do a great job.
- 31:59And it's often difficult to get to that point, right?
- 32:02So just a lot like talking about the price and the value together
- 32:06and when do I pay for it? That's my biggest complaint.
- 32:08OK, awesome. All right.
- 32:10Last question for the both of you before we wrap.
- 32:12If you were not building Konsugi at this moment, what would you
- 32:15be doing? Jeff, you want to go first.
- 32:17Well, I don't want to give all my great idea, all my great
- 32:19startup ideas away. Yeah.
- 32:22So I would, I would look at the say, I would look for the same
- 32:24thing. I would look for a boring
- 32:25business problem with a lack of innovation, something that's
- 32:28data heavy, something that people haven't innovative in,
- 32:31right. Maybe that's something like, you
- 32:33know, procurement technology, maybe that's something, you
- 32:36know, and reporting. There's going to be something
- 32:38out there that's a problem that nobody wants to solve, but
- 32:41there's a lot of money. And so I go find the problem.
- 32:43I think compliance is a place you can find a lot of those
- 32:45problems. Finance is a place you can find
- 32:47those in sales technology and marketing technology.
- 32:51I think there's a lot of great things that can be done.
- 32:53So I'd go find another important problem with a giant market, and
- 32:56I try to change the world with it.
- 32:57Basically, you are glutton for punishment.
- 33:00Yeah, apparently, yes. OK, Pingen, what about you?
- 33:04I think we both are and Jeff kind of said what I was going to
- 33:08say, but I would say I would still be building startups with
- 33:11the same people. I think oftentimes people don't
- 33:15realize that it's not the idea that fails, but really when the
- 33:18cracks start to show is when it gets tough with your founders
- 33:21and like your like leadership team.
- 33:24We have a phenomenal group of people starting with Jeff to
- 33:28Chris, our head of engineering, to Lexi, our head of sales to
- 33:31steel, our head of marketing, Ronnie, our head of CS lies our
- 33:36head of tax. And really like, I think when
- 33:39you find your core group of people who you live with and
- 33:42that's that's on to you, right? You have to like, it's almost
- 33:45like dating. You go on multiple dates, you
- 33:47find the person that's fixed with you.
- 33:48But when you do find that the life becomes goes from like
- 33:52being on a hard mode to going on an easy mode.
- 33:54So I would say like I would follow the exact same process as
- 33:57Jeff because I'm also like looking for public punishment
- 34:00and find that like, you know, selling shovels during a gold
- 34:04rush, as I'd like to say, but with an extra emphasis of when
- 34:09you find the people who you want to build companies with, that's
- 34:12what you do. Because at the end of the day,
- 34:14that that's what matters. We all can code, we all can make
- 34:18things, but it's appreciating that part of journey and
- 34:22trusting that other individual, right?
- 34:24Because start-ups is the highs of highs and the lows of lows.
- 34:27Somebody that you want to battle with, basically.
- 34:29Yeah, yeah, exactly. Finding your crap?
- 34:32And I think like it can suge out of the, you know, 130 people, I
- 34:36want to say like 70 to 75% of them.
- 34:39We've worked one of so many of the companies worked with in the
- 34:41past. So like a lot of the teams that
- 34:44I've managed and the people I've managed over the years now work
- 34:47for me again and the same for everybody.
- 34:49And so I think that's also, you know, it's a philosophy.
- 34:52I think you want to be, if you know, you're building the right
- 34:54thing, you just need the right people.
- 34:55And those are the, and you probably already know those
- 34:57people if you've, if you've been out there and you've had careers
- 35:00and you've built relationships and things like that.
- 35:02Yeah, guys, this was amazing. Thank you so much for joining
- 35:05me. Is there anything else that you
- 35:06want to leave the audience with before we close up here?
- 35:08What I would say is I think I'll do a shameless plug sales taxes
- 35:15is is one of those things where you might not be top of mind for
- 35:19you because you have 99 other problems, right?
- 35:22You, you're trying to figure out product market fit and you're
- 35:26trying to figure out your ads, your marketing, your sales and
- 35:29suddenly your revenue takes off. And then you have other problems
- 35:33that you need to stop off your onboarding, your CS, you need to
- 35:36scale things. But if you do not get your
- 35:41foundation right, it will like sales tax and compliance is one
- 35:45of those things that would always come invite you.
- 35:48I always remember this quote from some folks and people often
- 35:52forget they're like, oh, because I'm building an EI native thing.
- 35:55I don't I'm I'm somehow protected from this, but first
- 35:59time founders, they focus on product market fit and they
- 36:02focus on product experience Founders, they focus on setting
- 36:06up the right foundation so that they can scale and that
- 36:10principled approach is much, much more important.
- 36:12So really understanding your how does your company make money?
- 36:16Are you compliant or not? Are you running a structured
- 36:18business? I think that is something that I
- 36:21would say everyone should like, really think about and sit with
- 36:24Jeff you'd. Yeah, I think on on my side, I
- 36:27would say just, I think there's so many different tools and
- 36:31technologies, but actually spending the time to get good at
- 36:34them, I think like it takes, it could take years to get good at
- 36:37using the tools that are already great.
- 36:39And so like that's something I kind of see day-to-day is
- 36:41actually spending the time to do whatever you're using.
- 36:44I think people pay a lot for software and they have a lot of
- 36:47tools and I don't think they get everything out of them They can.
- 36:49I think a lot of times that leads them to go and buy a
- 36:51second tool or a third tool they don't need or maybe not
- 36:54understand like the value of the tool that that they have.
- 36:56So like just when you're looking at, you know, it's building, you
- 37:00know, building product today, we get so many great tools and
- 37:03products and it's like being a kid in a candy store using them.
- 37:07But I think it actually takes some effort to try to use them.
- 37:10Things are not as simple as they used to be.
- 37:11There's a lot more nuance about the right approach to to take.
- 37:14Awesome. All right, Jeff Pujan, thank you
- 37:16so much for joining us today. And to all our listeners, if you
- 37:20like this content, please give us a like, drop us a comment and
- 37:23let us know so we can give you more.
- 37:25Thank you and see you next time.