Latest / State of the AI Union / Michael Hoy: Show Me the (AI) Money
Transcript
- 0:03Hi everybody and welcome to this week's episode of the State of
- 0:06the AI Union. I am your host Laura.
- 0:09Now this episode is part of our startup series where we sit down
- 0:11with people who are building the next generation of AI native
- 0:14companies. Today I'm joined by Michael
- 0:16Hoyt, CEO and Co founder of Atlas, the only platform
- 0:19connecting real time monetization with building
- 0:22control. Michael is a three time founder
- 0:24and former VP of Sales at Pendo. So Michael, welcome.
- 0:27I'd love it if you want to introduce yourself and also
- 0:30Atlas. Yeah, thanks for having me,
- 0:33Laura. Really appreciate it and you did
- 0:34a great job introducing us. Not certain I I would add
- 0:38anything more other than if you're an AI company and you are
- 0:43just getting started or in your Series A Series B and you're
- 0:46trying to figure out your pricing and packaging.
- 0:48Our product is great for not just learning what business
- 0:52models work best for you, but deploying them rapidly.
- 0:55No engineers. We install through an SDK that
- 0:59allows us to be able to collect all of this insight for you and
- 1:03change your business model on the fly.
- 1:05Awesome, and Michael is also I just heard a former band member.
- 1:09He plays the bass guitar. Yes, yeah, that was a past life.
- 1:13My my touring musician past life.
- 1:15Now I have to wear a bunch of sweatshirts to cover up my
- 1:19tattoos. There you go.
- 1:22Okay, fantastic. So Michael, we have been talking
- 1:25a lot about AI innovation on this podcast series.
- 1:29We talked a lot about, you know, what customers are looking for,
- 1:32et cetera, but we haven't talked too much about monetization.
- 1:36And so tell us what your thoughts are on the shift from
- 1:40innovation to monetization, and what founders should start
- 1:43thinking about as they're building their AI native
- 1:45companies. It's a tough question,
- 1:47especially since we're in a we're in a time frame where what
- 1:51used to work doesn't work anymore and yet there's very
- 1:54little historical data to move from.
- 1:57So it's very much pricing and packaging.
- 2:00Business modeling in the AI world is very much the Wild West
- 2:03right now, hence why it's so critical to be able to test and
- 2:06iterate very quickly. But the reason why we're we are
- 2:09where we are. Obviously AI is bringing such
- 2:12new influence into the space. There's so many more things we
- 2:15can do with it. It provides very different value
- 2:18than the SAS technology we've used in the past.
- 2:22And because of that, the SAS business models that we've used
- 2:26in the past don't work. The two most common ones, seat
- 2:29based pricing and access based pricing really 'cause AI
- 2:34builders a lot of problems. So yeah, but most people are
- 2:37building AI products to make companies more efficient.
- 2:42And so a seat based model that kind of cannibalizes the value
- 2:46there and no longer no longer serves for great scalability for
- 2:51for an AI product that's trying to come in and maybe reduce
- 2:53heads or improve efficiency. And an access based model I
- 2:57would say doesn't, doesn't never work.
- 3:00That's an awful way of saying it, but it, it also isn't the
- 3:03best because we've got this brand new cost.
- 3:06AI is expensive to deliver almost sometimes to the tune of
- 3:10like 30 to 40% more in margin reduction than than our SAS
- 3:15products in the past. And so therefore, just getting
- 3:17every button, all you can eat of your platform puts you at major
- 3:22risk for, you know, going underwater on specific accounts,
- 3:27especially if they are overusing your product.
- 3:29And so. So anyway, we're at.
- 3:31We're at a a a, a crazy inflection point of and, and a
- 3:36great opportunity to test and iterate and try to figure out
- 3:39what works here. Yeah.
- 3:41And you know, you talked about the two versions of traditional
- 3:45SAS pricing, seat and access, right?
- 3:47And AI. We've been talking a lot about
- 3:50like value based pricing with AI or consumption based pricing.
- 3:53And I think especially with a value based pricing, that value
- 3:57can change depending on like the use case and the model update
- 4:01and the data set and blah, blah, blah.
- 4:03And there's a lot of like unknown and maybe like also like
- 4:07uncapped, you know, So for, for the, for our startups, you'll be
- 4:11like, well, am I cannibalizing if I don't price it right?
- 4:13And then for the buyer, they're like, well, if we agree to this,
- 4:17does it mean that's going to be like huge cost?
- 4:19Right. Right, that's more about about
- 4:21how the dynamics working out. You know, so, so, so well said
- 4:25that that uncertainty is really difficult for founders to
- 4:29navigate right now holistically. Our guidance to most very young
- 4:36founders, and I don't mean young in terms of their age, I mean in
- 4:39terms of their businesses age is if you're, you know, sub $1
- 4:44million right now, pricing and packaging should be as simple as
- 4:49possible. Your goal is just to prove
- 4:52someone will pay to solve the problem that you're building for
- 4:56in the way that you are building to solve it.
- 4:59That's it. It's very, very simple.
- 5:01And actually, like there's a great influencer, his name's
- 5:03Kyle Poyer. He runs a content rag called
- 5:07Growth Unhinged and he's got a great way of thinking about the
- 5:12role and not just the importance, but the role and the
- 5:14responsibility of pricing and packaging as a company scales.
- 5:19And before really before your first million, like what you
- 5:22need to think about is just proving that people will pay to
- 5:25solve the problem you're solving in the way that you're solving
- 5:27it. After that it does become more
- 5:29complex, especially in the in the AI world because these, you
- 5:34know, because you can no longer revert back to these old models,
- 5:37just as you said, you know, value based, consumption based,
- 5:42you know, a lot of people will throw out outcome based, right.
- 5:46All of, yeah, all of these models are options.
- 5:49But to your point, the when you use them varies not just, you
- 5:52know, not just based on, you know, the the product you're
- 5:55delivering, but maybe even all the way down to the customer
- 5:58you're delivering it to. We had this challenge at Pendo
- 6:01where our product drove very different value to very
- 6:06different customer verticals. We would, we were pretty
- 6:10vertical agnostic. We sold to anyone making
- 6:13software, creating products and our software would drive 10X the
- 6:18Roy in a fintech business than it would an edtech company just
- 6:23strictly because of the way they monetize their own product and
- 6:28and how people used it. But we had to use the same
- 6:31business model for both of them, you know, and so the
- 6:35acknowledgement really came down to the time of transaction where
- 6:39we discount one much lower than we would discount the other.
- 6:43I think now founders and builders have the opportunity to
- 6:47explore this a little bit better, understand their
- 6:50customers a little bit more, and specifically around what value
- 6:56does my product drive into that specific business, How can I
- 7:00potentially quantify that? And then how do I build a
- 7:05business model maybe for a specific customer, customer
- 7:08segment that extracts as much wallet share as possible while
- 7:13also meeting the customer exactly where they are getting
- 7:17the value right at that at that equilibrium of value exchange.
- 7:21So it's one of the reasons why I love AII actually think we've
- 7:23got the opportunity to to meet our customers exactly where they
- 7:27are in terms of the value that they get here.
- 7:30It just it's going to take a little bit of learning.
- 7:33Yeah. And so like I'm hearing you say
- 7:36like, OK, if you're sub million, you go to the customer and
- 7:39you're like, well, how much would you pay for this?
- 7:42And, and they'd be like 5, OK, fine, whatever.
- 7:45You know, like let's start there.
- 7:47And then afterwards, you just sort of like decide how much
- 7:51value you're bringing to the customer and then how much of
- 7:54that. Yeah.
- 7:55I, I, I think early on, it's regardless of the type of
- 7:58product you're building early on, you really need customers to
- 8:02use it to be able to prove and show how you know what the value
- 8:06of your product is. I think sometimes it can be
- 8:08real. It can keep us from moving as
- 8:10fast as we possibly can if we try to spend a lot of time
- 8:14triangulating the value our product will, will deliver
- 8:18without actually having any data around it.
- 8:21So I think it's not just, hey, we want to prove that customers
- 8:23will come in and pay to solve the problem that we're solving.
- 8:26It's also we need to go get real life ROI data to then be able to
- 8:32create not just a price point, but a package where customers
- 8:36feel high sentiment purchasing it and we're able to go get as
- 8:40much as much money for for that exchange as we possibly can.
- 8:43Let's talk about pricing and packaging.
- 8:44So you mentioned, OK, as simple as possible.
- 8:47So what kind of pricing and packaging have you seen that is
- 8:51reflective of this principle here?
- 8:53Yeah. So we, this is a, this is a good
- 8:55question because I think this is, this is the answer is it
- 8:58depends the classic consultant answer, it depends.
- 9:01And everybody's got I think really fun like 1st 10 customer
- 9:05stories in terms of the contracts that they've written
- 9:07form. We certainly do.
- 9:10You know, I think regardless of the type of product you're
- 9:12selling, really simple here like you can break the rules in your
- 9:16first 10 customers, 1st 20 customers, right?
- 9:18This is we, we, we broke the rules.
- 9:21We, we did an access quote UN quote based pricing model for
- 9:26our, for our first 10 customers. Now we did communicate to our,
- 9:30you know, actually our first 50 customers.
- 9:32Hey, our, our intention is to add an outcome based pricing
- 9:37lever here, which because we're billing infrastructure can be
- 9:40very Simply put as like transaction fees, those are
- 9:43coming, but you're our first 50 customers.
- 9:45So the for this, for the lifetime of this contract or the
- 9:48term of this contract, you don't have to worry about that, but it
- 9:52can be as simple as, you know, like I said, breaking the rules
- 9:55doing access based pricing. I think one of the things that's
- 9:58really cool that I've seen some startups do is credit based draw
- 10:02downs. Now, this is a very polarizing
- 10:05business model in the space right now.
- 10:07But I think if you're just getting started and you're
- 10:10trying to prove the value of your product and you're trying
- 10:13to quantify that, but you want your users to put some skin in
- 10:17the game. Allowing customers to come in
- 10:19and credit card swipe 20 bucks of credit in.
- 10:22There is a great way of not just getting users to put skin in the
- 10:26game, but also seeing how they spend and deploy those credits,
- 10:30right. You know, there's, there's no
- 10:32better proxy for value than actual value, right?
- 10:36And I think a great example of this are like the Lovables and
- 10:40the cursors and the Windsurfs of the world where they have a very
- 10:43prosumer type introduction to their product, right?
- 10:48And that can be very simple. You know, 20 bucks, $25.00 is a
- 10:53simple price point, right? Depends on what your product is,
- 10:59because if you were selling to B to B customers who are high in
- 11:03the enterprise, you might not do something like that.
- 11:06You might do something more along the lines of hey, we'd
- 11:08like to do a paid pilot. It's going to be a flat fee AKPI
- 11:13based that sort of thing. But I, I think for those first,
- 11:17you know, 1020 customers, it's OK to break some of these rules
- 11:21and keeping it very simple is, is the goal.
- 11:23And I like what you said about, you know, the credit drawdown
- 11:26model is a way for you to actually put some skin in the
- 11:29game as the customer are true, right?
- 11:30Because then you are forced to use it and see whether or not
- 11:33you you're finding value because seat based pricing is like,
- 11:36yeah, I pay for the seat, but are you doing anything?
- 11:38You know, so I think you're absolutely right there.
- 11:41So my goal, we have been talking a lot with AI companies and help
- 11:46us understand how Atlas solves for these problems on pricing
- 11:52and packaging that these new AI companies are, you know, trying
- 11:56to build for and stuff. Yeah, Yeah.
- 11:57So I mentioned earlier that not only do we need to find new
- 12:03business models, but there's not a lot of historical data to move
- 12:07from. And therefore, right now, one of
- 12:10the most important things we can do, we being like, you know,
- 12:14founders leading our businesses trying to learn what is going to
- 12:18work best. One of the best things we can do
- 12:20is start to collect as much data as we possibly can in order to
- 12:25make good decisions around our business model.
- 12:27And so one of the one of the critical things that Atlas does
- 12:31is it soaks in all of the usage of your product or your agent
- 12:37very similar to a product analytic solution.
- 12:39In fact, half of half of our team comes from a great company
- 12:43called Pendo where we really cut our teeth serving product
- 12:47people. And and so you can kind of think
- 12:49of part of what we built at Atlas being a, a technical
- 12:53solution that collects usage very similar to like a product
- 12:56analytics tool. Now what we do with that is very
- 12:59different than what a product analytics tool would do.
- 13:02First, we are able to create invoices out of that data.
- 13:06So if you do a usage based model or a consumption based model or
- 13:11you wanna do a credit drawdown, we take that data and very
- 13:14accurately depict that on an invoice based on what you and
- 13:18the customer have decided. The entitlement is the second
- 13:22piece of what we do, which is kind of similar to a product
- 13:24analytics tools. We surface all of that
- 13:27information in insight form and proactive suggestion.
- 13:32Now we do that through the lens of entitlement.
- 13:34And what that means is we look at how people use your product
- 13:38based on what they've purchased. We believe outside of value for
- 13:41value being a great proxy, usage is a great proxy to value.
- 13:45And so if you're unsure of how to price and package your
- 13:48product, one of the first places you should look is how do people
- 13:52use it? Where do they get value out of
- 13:54it? Right?
- 13:55No, it's just a single data point in a very large discussion
- 13:59around what the business model should be.
- 14:01But we try to bring not just that data point, but perspective
- 14:07suggestion to you around what those data points mean and where
- 14:10else to look to start trying to hone in what the best business
- 14:14model might be. The last thing we do is because
- 14:18we install via this SDK, because customer, your customers are
- 14:23coming in and buying entitlements and doing contracts
- 14:27through Atlas, we're able to then provision the product based
- 14:30on what they've purchased. If you're running a consumption
- 14:33based model and you've got a maximum level that that that
- 14:37that company or that user can hit in terms of usage, we're
- 14:40able to actually block or pay wall that customer from using
- 14:44more right And some or we're able to surface different parts
- 14:47of the product based on what the customer has bought.
- 14:50Some people think this is somewhat prohibitive.
- 14:54But if if you're ACEO buying a usage based product, knowing
- 14:59that your team can't go spend half the year's budget just
- 15:04crushing, you know, an LLM is is is very comforted.
- 15:09So, so I think, I think it's really important that, that
- 15:14whenever you do business models like this, you have some of
- 15:16these guardrails. And so Atlas helps you, helps
- 15:19you do that. So if you're going to search for
- 15:21pricing, market fit, which is a, a term we're coining, you need
- 15:25to be able to test and iterate really quickly, right?
- 15:28And very much like what you, what you would do with your
- 15:32product. And because Atlas does these
- 15:36things and you don't need an engineer to make any changes
- 15:39once it's in, you're able to deploy a business model, see how
- 15:42it works, see how it affects the usage of the product.
- 15:46Understand through, you know, Rai, what changes you can make
- 15:50to better convert or improve the, the sentiment or the upsell
- 15:55opportunity of a certain package, and then quickly go and
- 15:58make more changes without actually going into your code to
- 16:02do it. And this allows teams to iterate
- 16:04very, very quickly on their business model and try to find
- 16:07the the right mechanisms for either a customer or an entire
- 16:11customer base. That's awesome.
- 16:13So you're, you're primarily a solution for vendors that are
- 16:17looking to help, you know, control monetization and stuff
- 16:20like that for their customers, but you also have a customer
- 16:23solution that helps customers value manage their AI solution.
- 16:30Yeah, yeah. Like, you know, I think, I think
- 16:33that's one way to think about it if we're on the same page,
- 16:36right? You're you're, you're thinking
- 16:38more along the lines of how do I understand what the actual value
- 16:42that my agent is? Yes, exactly.
- 16:45Yeah, I think now granted this is an area where we want to do a
- 16:49lot more work, OK. But initially, you know, there's
- 16:55a like I mentioned, like the one of the better, one of the best
- 16:59proxies to value is usage. And and we can certainly go in
- 17:03and help you understand not just how the agent is being used,
- 17:08but, you know, the beginnings of what that value is now.
- 17:12Holy cow. Like I can't wait till we get to
- 17:14the point where we can resurface back to the agent builder or the
- 17:19agent owner. Here's here's the ROI you drove
- 17:23your customer today. We're not at that point yet, I
- 17:26think. Yeah, Yeah.
- 17:29That is that is futuristic. But transparently it's not.
- 17:31It's not that far in the future. Awesome.
- 17:34It's not that far in the future for us, certainly something that
- 17:37that we have spent a lot of time thinking about.
- 17:40My Co founder Clay Minor was the SVP of post sale value team and
- 17:47so very near and dear not just to his heart, but also to our
- 17:51understanding of of you know, the importance of of sticky
- 17:56software inside of a business is understanding what the ROI is.
- 18:01And so much of that work that we did in building value frameworks
- 18:07and understanding the ROI Pendo drove into other companies.
- 18:11We're taking that and trying to create really great product
- 18:15around that, just as you said. So merchants understand the ROI
- 18:19that they're driving to their customers and can manage the
- 18:22customer base based on. Awesome, very cool.
- 18:24So we did talk about how Atlas is the only platform right now
- 18:29connecting monetization and billing control.
- 18:31But I did want to ask you, in such a crowded marketplace,
- 18:35especially with AI in every deck containing the word AI, how do
- 18:39you differentiate yourself and or what advice would you give to
- 18:42founders that are building an AI?
- 18:43We have spent the better part of October talking about this.
- 18:49Hey, irrelevant. Oh my gosh.
- 18:52And, and, and I should say I, I hope that most are, are spending
- 18:57time on this. This is just part of the world
- 19:00we live in right now is you're no longer going to have 5
- 19:03competitors. You, you know, you're probably
- 19:06closer to having 50 than 5, which is a, a great thing.
- 19:09Software has become easy to create.
- 19:12That's awesome. You know, we've really
- 19:14democratized, you know, the creation of, of software And I,
- 19:18I think that's on the whole a good thing.
- 19:21I think right now it's, it's really challenging for founders
- 19:24because there is a lot of noise and it's really tough to, to
- 19:29find signal. And in fact, we're, when we talk
- 19:31to investors, this is one of the big questions that they ask is
- 19:34like how, how do you get Atlas through all of this noise?
- 19:40You, I can tell you what we've come to, but like it might be
- 19:43more impactful to, for, for the founders listening to share like
- 19:46how we, how we got here in the process we went towards, you
- 19:51know, just figuring out how could we differentiate in the
- 19:54space. We, we did a lot of exercises,
- 19:58you know, first and foremost, we pulled all of our competitors
- 20:01copy off the web and we stuck it next, you know, next to him.
- 20:06We tried to find gaps and holes in terms of things they didn't
- 20:09talk about, terms they didn't use, concepts that they didn't.
- 20:13We also lined out our understanding based on our
- 20:17customers feedback what the two to three most valuable things
- 20:22inside of our product are. We also asked the same question
- 20:26what are the two to three most valuable things you want us to
- 20:28build? And we, we started to
- 20:31triangulate between what are our competitors not saying, what do
- 20:36our customers use about or think about us that's most valuable.
- 20:39And then like, what is their vision?
- 20:41How do they really think about Atlas in the future?
- 20:46And you know, we would then put our own little spin on it.
- 20:49And we tried to find these areas that were untouched, these words
- 20:54that weren't said, but we tried to do it very much in the
- 20:58language of our customers. And so by no means do I think
- 21:01we've gotten this right just yet.
- 21:04It's an iterative process. But what I would encourage is
- 21:09don't throw this one into GPT. Try to do this one as human as
- 21:15possible. One of the things we noticed
- 21:17putting all of our competitors messaging next to each other,
- 21:22you know, and honestly next to ours after we had GPT D, ours
- 21:25was, Oh my gosh, it's all the same voice.
- 21:27A. Lot of the same words.
- 21:29It's a lot, you know, OK. And we, we kind of came to this
- 21:31conclusion that, shoot, we could just be different just by typing
- 21:35something up ourselves, which we ended up doing.
- 21:38Yeah. You know, I think it's, I think
- 21:39it's really challenging regardless.
- 21:41Like, you know, I, I make it sound like that's an, that's an
- 21:44easy solve. We, we have not figured this
- 21:46out, though we are, we do think that what we have right now is
- 21:51different. But I, I was talking to another
- 21:53founder about this the other day.
- 21:55I actually think to cut through this noise, we have to create
- 21:59more of an experience. The words matter 100%.
- 22:04But your digital real estate's really important.
- 22:07And when someone drops onto any part of your digital world, what
- 22:12is that experience going to be like that differentiates you
- 22:15Now? I don't, I don't know what that
- 22:16is. Maybe it's a maybe it's a a
- 22:19voice, you know, voice activated.
- 22:22Welcome. Maybe it's, you know, maybe it's
- 22:25maybe it's a oh, you know, I've got a great, great founder buddy
- 22:30who built a company called Asset Mule.
- 22:32Maybe it's maybe the website's super interactive and, and it
- 22:36allows the customer to be able to build their own sales assets
- 22:39to take back to their team upon evaluation, right?
- 22:42Maybe it's, maybe it's something, but I, I, I think the
- 22:45words, the words are going to matter, but maybe matter a
- 22:48little bit less than the actual experience to help kind of cut
- 22:51through the noise a little bit and help people understand
- 22:53really who you are and what makes you as a business or you
- 22:57as a software or you as a team different.
- 23:00With all this noise. The pendulum has swung over here
- 23:03and we will start to swing back towards the middle where we'll
- 23:06find really great ways to differentiate.
- 23:09Yeah, and it's almost like, don't confuse speed and
- 23:12efficiency with effectivity, right?
- 23:16Oh. My gosh, yeah, there's a time
- 23:18and a place, right? But I think we've we've
- 23:20definitely opted for speed always, right?
- 23:24Sometimes it's really valuable to slow down, to speed up.
- 23:27And Michael being a three time founder, is there anything
- 23:30different about building an AI native company this time that
- 23:33has made it on giving you some, you know, much really different
- 23:36learnings? Yes, yes, yes, yes, yes.
- 23:40I'm having trouble whittling that down to a very simple.
- 23:45So I started so, so my first two companies I, I started within
- 23:50six years of each other, you know, and that was 15 years ago.
- 23:57So, you know, times are very different then.
- 24:00And, and I was fortunate enough to do almost a 10 year stint at
- 24:05Pendo in between and really learn how to build great
- 24:08companies, great cultures, great teams within a a large, rapidly
- 24:12growing organization. And I can tell you that of
- 24:15everything I learned, I will, I will still use some of it, but
- 24:18very little has prepared me for building a company in, in the AI
- 24:24space. And I, what I, what I anchor on
- 24:26is, well, no one's really done this before.
- 24:29Again, there's not a lot of historical data to show the best
- 24:32way of doing this right. And I think you'll hear a lot of
- 24:35stories around when is the first billion dollar single person
- 24:39company going to happen? Right?
- 24:42When is when is the first billion dollar ten person
- 24:44company going to happen? I still think that these are
- 24:48very far off if, if ever, but it, it certainly pushes you to
- 24:54challenge like a lot of the things that I'll say, like we
- 24:57grew up professionally knowing, right, or, or believing you want
- 25:03to increase sales numbers, hire more sales people, right?
- 25:07You got tons of customers to serve, hire more CS people, got
- 25:10more code to write. Well, shoot, get more engineers.
- 25:13You know, I think I think this is a good challenge for us as
- 25:17humans because I think we got very stuck in, in the thought
- 25:21process of, well, there's only one way to solve the problem.
- 25:23Let's throw heads at it right now we're we're trying to swing
- 25:26the pendulum in the complete opposite direction.
- 25:28Well, there's only one way to solve this problem.
- 25:30Hire an agent. The the reality will fall
- 25:32somewhere in the middle. And what I am really excited for
- 25:36is the first billion dollar company that has 150 people, 200
- 25:42people, and each one of them manages a team of agents that
- 25:48that makes them superhuman. And so where this becomes very
- 25:55different from the companies that I founded in the past is we
- 25:59had plans for how we would grow those companies.
- 26:02And the plan for how we grow Atlas looks totally different.
- 26:05You know, we, we one, want to continue to rapidly grow, you
- 26:11know, 200, three, 100% year over year.
- 26:14But we want to be able to do that while not hiring, you know,
- 26:19by hiring 1/4 of the humans that we originally thought we would,
- 26:23right, right. We, we need to pay so much more
- 26:26attention to our margins to be able to do that, to our
- 26:31efficiency to be able to do that.
- 26:32We measure EBITDA, which was not something I learned about until
- 26:35my like 8th year at Pendo. And but it doesn't, it hasn't
- 26:40made us any less ambitious. I could just has changed the way
- 26:44we think about building the company.
- 26:47And I just got out of this conversation with my head of
- 26:50operations, who's one of her main roles next year is going to
- 26:55be setting up the infrastructure to build a company of agents.
- 27:01Now, we're always going to have people and we need incredible
- 27:04people, humans who are super intelligent, who are both like
- 27:09high-powered ICS and managers. And what they'll do a lot of
- 27:14management of is their team of agents and we need to be a 15
- 27:18person company that acts like 100 person company.
- 27:21And so that's not something we ever had the opportunity to
- 27:25think about when I started companies before.
- 27:27And I think it's opened up a, you know, if you can execute on
- 27:30that, man, it opens up a lot of options for you.
- 27:32Awesome. All right, Michael, last
- 27:34question then we will do something interesting after
- 27:37this. Well, this has all been
- 27:39interesting, so yeah. Yeah, we'll do something more
- 27:41interesting. OK, sure.
- 27:45Last question is let's zoom out to two years.
- 27:49What does the future of AI monetization look like?
- 27:53Will we all need some kind of monetization control?
- 27:57Will we need more control? Tell us what you think.
- 27:59Yes, in two years I think optionality rules.
- 28:04I think you'll as a merchant have the option as to whether
- 28:08you want to let AI price and package your product for every
- 28:12customer comes in very similar to the way you know any airline
- 28:16does when you go to buy a ticket.
- 28:18Think you'll have that option. I don't think that's the way all
- 28:21software will be bought. I think they'll be options for
- 28:24merchants to allow their customers to do what we call
- 28:27software made to order, where they really pick the
- 28:30functionality and the features or the the levels of usage that
- 28:35they get and they customize their own software packages.
- 28:38I still think a lot of software is going to be sold the way we
- 28:41see it today, you know, product LED, human assisted, you know,
- 28:45might have different models off of it.
- 28:48I do do believe that the hybrid model will reign supreme.
- 28:52Some sort of, you know, base fee plus something variable and the,
- 28:58you know, maybe for a different day.
- 29:00Have a big thesis around how we're going to hire AI like
- 29:03labor and pay it like labor, like sales people actually.
- 29:07But I also believe that a lot of the monetization process and why
- 29:12everyone's going to need a monetization mission controller,
- 29:16which is really a kind of how we think of Atlas and the single
- 29:18person that that will run all of Mission Control for that
- 29:22business. If there's a lot of human work
- 29:24that goes into just doing a single deal right now, obviously
- 29:29human work to bring deals into the pipeline, lot of human work
- 29:33to administer the admin and keep all the other humans up to date
- 29:37as to what's happening. Once that deal is agreed upon,
- 29:41there's a lot of human work that goes into creating those
- 29:44contracts, completing those contracts, provisioning that
- 29:48product, recognizing what has gone into that contract and a
- 29:52making sense of it on the accounts receivable side.
- 29:55Like there's a whole lot of, and then of course, the, the
- 29:59recurring billing and tracking and, and there's a lot there
- 30:02that is all ripe for automation. And really what we, what we want
- 30:10at Atlas to push people towards is this idea that sales will
- 30:15always have some sort of human function to it.
- 30:19Yeah, we, we're emotional beings and buying things is inherently
- 30:25emotional. And So what, what we want to do
- 30:28is we want to take all of these processes that are not the
- 30:32emotional side of buying and selling.
- 30:34We'll move that to AI so that companies can focus primarily on
- 30:39serving their customers. Value is is emotional, it's
- 30:44perceived value often. And we believe the more that
- 30:48that a human can be with a human in that process and not with a
- 30:52spreadsheet or a piece of software rectifying bills or
- 30:56tracking usage, the more time they'll have for, for being with
- 30:59their, their human customers. And so that's what we're aiming
- 31:04for in two years is more and more automation of the
- 31:07monetization process that doesn't involve 2 humans
- 31:11talking. Yeah, I look forward to it.
- 31:14OK. We are going to play a version
- 31:16of Jeopardy which is called what is what is.
- 31:20OK, you ready? Yeah, we watch Jeopardy almost
- 31:23nightly in this home, OK? OK, what is an AI pricing model
- 31:31you've never seen but would like someone to try?
- 31:33Good question that we haven't seen.
- 31:36We have a customer who's ideating on this.
- 31:39I won't share this. I won't share who it is because
- 31:41it's it's not public. They might do it.
- 31:43They've got a very, very large enterprise customer of their
- 31:47own. The name rhymes with force.
- 31:52They want to allow their customer to use their product
- 31:57for free and then every time they catch a vulnerability
- 32:01charge Salesforce $1,000,000. And so, yeah, very extreme.
- 32:06The early, the early feedback from from their customer is
- 32:10maybe, maybe, yeah. Yeah, how, how, how do they feel
- 32:14about their vulnerability, you know?
- 32:16Yeah, Yeah. Well, I think the value that
- 32:18that custom that that product drives inside of Salesforce is
- 32:21very big, yeah. Awesome.
- 32:24OK, next question, what is 1 startup expense?
- 32:28You'll never cut Pizza Fridays. Pizza Fridays, man, dessert the
- 32:34OK, can I do my top 3? Is that OK?
- 32:37I always have trouble getting down to 1.
- 32:39So we're a remote team and and we we're all over the globe.
- 32:44We're everywhere. Once a year we all come together
- 32:47in person. And so this is yo, it's actually
- 32:50not just one expense that has many expenses to get everybody
- 32:53in person, but that is your priority number one.
- 32:57Priority number 2 is Atlas. I would not, I would not.
- 33:01We use our own product. I guess fortunately, well,
- 33:04technically Clay pays for it on his credit card, but I would not
- 33:07cut our product. And then honestly, like products
- 33:09like dev Rev that help help keep, you know, we're we're
- 33:13starting to explore yours. I hope, I hope you don't mind
- 33:15the you know, the the unsolicited, yeah the
- 33:18unsolicited commercial here. But transparently like we we've
- 33:22looked at Dev Rev for a lot of what you and I have just
- 33:25discussed is OK, how do you know our our head of operations is,
- 33:29is leading this. How do we how do we build any
- 33:32process inside of our business in an automated way, right.
- 33:36If we do something more than twice and how do we take
- 33:39everything out that isn't too human stalking, right?
- 33:43How do we automate that? And so, so those for us when we
- 33:48think about scaling that's critical and and as an expense
- 33:51we can't give up. Awesome.
- 33:53OK last question, what is 1 tattoo that you're hiding right
- 33:57now that you regret? None.
- 33:59No regrets. Wow, that's awesome.
- 34:01And that's a great thing, too. No regrets.
- 34:04No, now I've got a couple silly ones planned so ask me in in a
- 34:08few years. OK.
- 34:11All right. Michael, thank you so much for
- 34:13the conversation. Is there anything else that
- 34:14you'd like to tell the audience before we close out here?
- 34:17Oh my gosh, no, I just want to I just want to extend my, my
- 34:20thanks and my gratitude, not only like for, for today, but
- 34:23but the dev ref team has been awesome in, in partnering with
- 34:26us. Y'all are such good humans, such
- 34:28great people building. You know what looks like
- 34:30awesome, awesome technology. I can't wait to to spend more
- 34:33time with it and and all that. All that being said, I look
- 34:36forward to to a lot more shared opportunity in the future.
- 34:40Awesome. Thanks so much Michael.
- 34:42And to the our audience, please let us know that you liked it
- 34:45and like a comment and otherwise we will see you next time.
- 34:49Yeah. Thank you.
- 34:50This episode was brought to you by Devref for Startups.
- 34:52If you're a founder building with AI and have a story worth
- 34:55sharing, would love to hear from you at Startups at devref dot
- 34:59AI. Don't forget to explore the
- 35:01Devref for Startups program on devref dot AI slash startups
- 35:05where you can get up to 12 months of credits and much more
- 35:09to experience true team intelligence with computer your
- 35:13AI teammate.