How Apres CEO Matt Waite grew Apres to $3.5M revenue and 12 customers in 2024.

hello everybody my guest today is matt wait he's the ceo and founder of apre an automated automation platform for ai training they focus on eliminating the complexity and cost associated with training artificial intelligence applications their application lets anyone build their expertise into a training module to generate high quality data necessary to create and maintain intelligent software matt you ready to take it to the top let's do it all right man so this is a concept that sometimes it's difficult to kind of wrap your head around so make it simple for us what's the company do and how do you guys make money yeah absolutely so really what we focus on is making the data that teaches machines how to see how to hear and what to do so let's say that you had a machine that was trying to recognize stop signs well first it needs to recognize that there's a stop sign in the image so that requires a certain level of data on top of the image let's say and we provide that data so we do that at scale and we actually train machines to be able to produce that data more efficiently than anybody else even even more than like a mechanical turkey kind of model yeah so what mechanical dirk is doing is really throwing humans at the problem we take a learning approach so we're actually training an ai model to produce the data for your ai model interesting okay so who's paying for this so companies with large deployments of intelligent software or databases that need to be updated continuously i mean the real problem that these companies are running into is that as your model runs into the edges of what it knows it needs sort of to relearn what it doesn't know and so that's where they interact with the software like ours so that we can re-educate the model can you tell me a concrete story is there a kind of a customer story you can share on how they used you yeah sure so there's a company called reply ai and really what they do is they provide uh chatbots as a service to enterprise cosmopolitan hotel is a great example of one of their implementations and really um let's say the cosmo's throwing a a party on the weekend that's outside of the lexicon that they would normally use or the dialogue scripts that they would normally be able to reply to but what they do is they link us into the flow so when their ai doesn't recognize a conversation pattern it links us now if we don't recognize it or i rai doesn't then we actually route it to humans who can do the labeling the performance analysis necessary to pump it back into their model so that they can recognize that behavior wait sorry what's a chatbot have to do with with data on top of an image yeah so we do both uh object detection as well as textual data so not only can you you know from a textual standpoint understand language through ai you can you know you can understand visuals as well we solve for both use cases got it okay and then help me understand how you make money right so are you pure place sas company so it's actually based on the data that we produce for you so we charge a fee for machined insights so that's a very low fee a fraction of a penny and then anything that touches human hands has um you know sort of whole sense attached to it but we really only charge the data that we process so each query would be a um a single charge so i'm coming you today to kind of sign up i'm another version of the cosmo uh what like am i going to try and guess how many queries you're going to do over a year and that's going to be my contract value to you or do you bill me at the end of each month based off number of queries like used or something yeah so we actually act like a cloud provider in the sense that we're you know it's kind of a pay as you go model and that's really what um what they need it's because you can sort of walk crawl running in a sense but also there's very elastic needs and so charging a subscription fee really doesn't work um and typically we're dealing in the millions of data points it's really scale implementations yeah um get the most views so so really that's so help me understand just because we don't have time to go on every customer cohort but on average what would you say kind of an average kind of user of your platform is going to be paying you per month and about how many data queries is whatever that payment is yeah so large enterprise customers over the course of a year will pay you know around a million dollars and um you know a smaller sort of sas based series a stage uh scalable company would be paying you know five to ten thousand dollars a month okay do would kind of the five grand a month thing be more of a fair average across your entire user base or no yeah i would say that if it's if it's more of a b2b sale then yes that's that's sort of where we sit but we also sit with developers as well just because it's a pay-as-you-go product we you know are pretty accessible if someone's paying five grand uh put a put a query number behind that so 60 grand a year i'm going to be doing about how many queries uh 60 grand a year you're doing you know half a million million queries those those are the results you're delivering back to me right exactly interesting okay cool put this on a timeline for me when you guys launched the company about a year ago is when we kicked off the first line of code and we're actually about to do a public launch in a week or so okay so are you pre-revenue right now no no we have revenue oh it's just a function of the fact that we haven't told anybody really that we're on the internet where were you so give me context a year and a half ago i mean what were you doing what what made you jump into this idea yeah so i was running a financial services company who um and i was the cpo of the company i had to build a lot of processes to you know translate data between humans and machines and felt like scaling out that utility was going to be a net benefit for the market so i spun out of that company and started you know a whole new one that's great so 2018 you launched called a year and a half ago and then how many customers have you scaled to today so today we have about 12 customers that are active on the platform okay that's more than i thought you'd say i thought you're going to say maybe three four kind of you know individual cases but you have you've got 12 kind of pre-launch that's great yeah yeah and we've had a couple hundred businesses that we've interacted with over small implementations or small you know uh sort of interactions but yeah really like core good use cases i would say yeah around it doesn't and how did you get kind of the the first you know two paying of those 12 well this is kind of a uh a business where you can you know it's almost like a mechanical turk you really don't know what's happening necessarily and so um the data itself can be human engineered a lot of the time and so we just cut our margins until it made sense for the market to say yes to us and we ended up not making money for a long time built out the technology components that we could widen our margins and so having those conversations with people really with cash as our only differentiator initially is really what kicked off the first batch of customers and we were just very how'd they find you i'm talking about specifically how they find you so i uh had the idea for the company and put it on product hunt and i got a got the first cohort and then started reaching out to ai companies um just with a few you know good case studies um b2b you know typical repeatable revenue kind of how did you reach out like were you looking up linkedin kind of job titles of certain companies you wanted to target or how did you do it yeah so a combination of things um there's uh there's a great networking app called shaper and so i used a few various you know linkedin shape or a few other networking applications and just try to get in touch with as many viable prospects as possible just with the angle of let's talk about who are they though i'm trying to dive deep here what is the viable prospect like what's the title of the person you're trying to sell to inside of an organization ctos and uh anybody around data science okay got it to connect so you're looking for like the c2 at like the cosmo this yeah the cto of typically ai driven companies or that have obvious ai components so how do you quickly figure that out um based on their technology profile i mean if they're selling sas ai software anybody that you've ever interacted with um they have this component built into their business so it's a known problem interesting okay and um have you bootstrapped the company or did you decide to raise yeah no i funded it and um we got a small check at the end of last year but we're about to go out and raise that's great okay wait so how much in the company to date how much raised i don't want to talk about that but not a lot of money okay like less than 100 grand yeah okay fair enough and then why are i mean why can't i always root for people to stay bootstrapped because then you like you're like betting on yourself and i think it's just lazy to raise so like why do you have to raise why can't you keep bootstrapping so this this is a it's a rich industry but we really we want to make the platform basically um we want to continue with this pricing model that i just described to it allows for the largest um level of accessibility it makes the largest impact on the market and it allows us to go fast and in order to do that we kind of need to subsidize our cash flow gaps because like i just explained to you there's an elastic market here and so for elastic user and so those gaps are just going to create you know cash situations that i don't want to have to burden and so are there gaps i mean shouldn't be pretty consistent if they have a thousand api calls or queries in the first month shouldn't you know if it works shouldn't it go up to like then you know ten thousand then a hundred thousand and a million month after month it should stack right so we kind of yeah so i mean that's the general idea but that's not how engineers work all the time so they train and then they go out and they see a live implementation then they gather enough data so they can come back and they can retrain now if we're tightly in the loop of course yeah it can scale but also at the same time our machine is learning their core use case to put the human element which is the rich element of the business um kind of out of business so the goal there is to depress those costs for the users that we can retain them um and so that's the core mechanic of the of the company so um it's just the behavior right now the the industry is so new and the behaviors are so nondescript that um you know we're expecting sort of like quarterly um batches and or monthly batches and still there's there's costs associated with um running these machines on a daily basis that are non-trivial so you just kind of you need to be able to we need we need some cash to be able to front that now i'm not talking about going out and raising some illustrious valley what's the right amount for you to raise like this around what are you going to target i mean like a million or 10 million or two or what yeah one to two million is really what we're targeting an extremely modest amount from like a high technology standpoint yeah yeah well yeah no i mean look i think it's great and and round up the team for me is this your kind of your baby or do you have a co-founders i have four co-founders so holy crap four so there's five of you totally the other three other co-founders so four four okay that's great okay that's a ton i mean how do you guys always make sure you're on the same page and if you guys are let's say you just split it you know 25 25 25 25 if two of you guys are on each side how do you break a tie i mean how do you make those decisions yeah fair i didn't start with co-founders so um i started by myself i found a really accomplished sort of seasoned engineer who i wanted to add to the team and he's just rolled into a full-time co-founder and we had a few people reach out to us along the way that were core contributors to the technologies that we were using and so we grabbed them as well so it's not an even equity split so those types of confrontations don't necessarily happen um but how do we make sure we're on the same page we're completely 100 remote team and that offers a lot of advantages and disadvantages but we manage everything through you know sort of the stack that you would imagine and we just have sort of a rigorous way of communicating yeah i know i think that's great well and and so what do you i mean just to summarize some of the data points you've already given right so so 12 you launched in 2018 uh kind of 12 paying customers you articulate earlier you have some that are you know could be in the millions right in terms of acvs depending on usage and then you said kind of minimums or more like five grand a month or sixty thousand bucks a year so at a minimum of five grand a month on 12 customers that would put you at like 60 grand a month right now in revenue is that accurate yeah that's roughly right and so with a team of four people i mean you're i assume you're profitable on sixty thousand dollars top line um i mean is that accurate are you still burning cash no no we're we're roughly profitable or break even breakeven's fine too right you're reinvesting so yeah i don't know man my heart just wants to like try and convince you to not raise capital because it sounds like a great idea and uh and you're obviously gonna get deluded if you do that it just strikes me as like your ability to find a million dollars i mean if you're already doing 60 dollars a month you know you can push a little bit for a couple more months get two million run right i mean if you do raise capital how will you actually use it to speed up your execution is it engineering hires or something else so we do need to be buffered with a you know maybe one or two heads from an engineering standpoint but i really want to build out um a solid sales team i mean we have a we have a good product we found sort of a fit and it's an enterprise sale so it takes a long time i need bodies to be able to handle that yeah and so that's that's really where i would be investing that's been all you today you've done all the sales yeah i do everything i mean i am an engineer i've built the product that everybody uses but um and everybody on the team could build the product top to bottom but yeah i i now i just talked i try to talk to customers and stay out of engineering and that's where my team wants me so do you let me ask you if you go raise a million and then you go out and kind of hire your first kind of sales person what kind of expectations do you have i assume you probably modeled some like guess of what a pro forma might look like in terms of like number of leads they need to close like per six months or how long it's going to take you to onboard them based off your own performance selling how do you model that stuff because you're i mean everyone does this at this stage and it's a critical stage so how have you done it yeah i mean i i come from sort of like a repeatable revenue system but i can't apply that system because it's an enterprise sale right and so you kind of have to be more account based about the way that you deal with the market so i mean is the question about how do i expect you know this first sales person to behave or what kind of sales person am i hiring no it's the expectation so let's say i'm a vc you're trying to raise from you really want to raise for me i'm saying listen matt i love you dude uh we're gonna spend the million and you're gonna say well we're gonna hire two sales people at a base of a hundred grand and we're paying them this commission structure and we believe that over the next six to twelve months as they scale it's going to take them four months start hitting their quota and you know each of them should be adding about half a million in arr per year by the end of 12 months like i'm just curious if you've thought through modeling that at all yeah i think that you pretty much hit it i mean it takes the average we're not gonna hire a hundred thousand dollar sales person we hire sort of bottom up so i mean it's gonna be more of like a uh an sdr ish but i mean you know 60 70 80 full in depending on the region that we end up hiring three to four months to to ramp and then yeah the expectation would be um i think we have it at 350 to 500 000 that's a pretty good guess on my part just pulling out of it a sales person kind of has to pay for themselves four or five times over right yeah yeah i think that you nailed it where is so maybe a question for you as a ceo how aggressive are you willing to be on fully weighted customer acquisition cost in other words would you pay full first year acv or what you think they'll pay you in the first 12 months to get the customer are you less aggressive so where i sit on that is we have to do a lot of times a lot of upfront work to get a customer and so there's there's not a lot of in-room conversation because the data kind of does the talking for you and so there's not a lot of at least it's my experience and the larger deals that i've handled it's really let the data let the data speak it's just expensive to get that data so i yeah i would go up to um at least our first three months of cost that's kind of my threshold okay um and that's that's kind of where i sit on that's good is that a five grand account a month that you say hey you'll spend up to 15 grand to grab them and then it's a three month payback that's you know cash gap that's manageable yeah exactly that's very cool too early i imagine to talk about things like churn right you probably don't have any churn no so churn isn't really uh something that we understand too well at this point yeah yeah and then um you ju when was your first paying customer signed up do you remember our first paying customer was january last year january of 2018. okay so where did you if you're at 60 000 a month today where were you exactly a year ago do you remember uh i think the i think the first rollout we had was between 500 and a thousand dollars okay a month a month yeah okay so in in may of 2018 so a year ago you basically had one customer paying a grand a month no that was in january so i mean by this time last year okay to put a uh i i think that we probably were making yeah between like two to three thousand dollars or something like that still fair enough and that's great then do you remember what you closed 2018 at so six months ago december 2018 yeah yeah i don't want to talk about that though if you don't mind no that's okay but i'm curious why you already told me 12 months ago you're three grand a month now you're 60 grand a month so i imagine you ended last year somewhere in the middle of three grand and 60 grand a month yeah it's it's okay um but yeah it was somewhere somewhere in the middle of that for sure wait i'm curious there's like a strategic you're a calculated guy there's a strategic reason you're not sharing that number i'm curious what it is don't show the number but what's the what are you thinking right now no it's just uh well because it'll tell i don't know it's it's it's i don't i don't like to share numbers that openly like the numbers that i've shared are sort of roundabout numbers as well so um no it's a uh it's it because it indicates really velocity and the types of customers that we're taking on and so i don't i think that if we're in a position array it's best to just keep that close to the vest yeah but just to be clear those numbers that you already told me i i was i wasn't i mean i wasn't wanting to put words in your mouth you told me 12. i know yeah yeah okay yeah you told you told me 12 customers and 5 grand a month minimum so you're doing at least 60 okay cool yes sure that's great good all right and then um and then um so okay so rounding this out um you're looking to raise right now um obviously driving growth healthy payback period trying to hire sales people with the money you raise um do you have i mean for the end of 2019 what's kind of the goal you have to end a 2019 is it a team based goal or you know a money raised base goal a customer a or based goal what is it yeah we um so our kpi is uh humans out of the equation so for every data point that we get we want to have humans 90 out of the labeling process and as you can imagine that affects all aspects of the business in a positive way and so that's really our kpi how do you measure that though uh it's it's actually pretty easy um so if you tell me i don't understand for sure for sure that's i'm getting there it's uh so 10 000 data points come in or let's use a you know 10 000 is fine 10 000 data points come in we want to be able to machine uh 9 000 of those so what that means is that rather than touching human hands it um it hits a machine and so rai would actually physically label the data for you and you'd only be paying for at that point a thousand hand-labeled data points so that's our kpi why is that the kpi i mean you can just slap any label on it how do you measure accuracy of the 9000 that you've removed from the human hands yeah so it still goes through a human process it still goes through like a level of overview and it still goes through a review process um but you know all of that is kind of done um uh probabilistically so it's not necessarily a hundred percent certain accurate 100 of the labels are 100 correct it's probabilistic which is why you're paying a thousandth of what you'd pay you know for a human label yeah all right very good let's wrap up here with the famous five number one what's your favorite business book my favorite business book um lean startup number two is there a ceo you're following or studying today i love elon musk number three how many what's your favorite online tool for building your company gitlab uh number four how many hours i sleep to get every night six and what's your situation about married single kiddos i have a kid and that explains the six all right so not married no i'm married oh mayor i was a second good as i said usually they come together okay married one kiddo and how old are you i'm 30 years old 30 years old all right last question what do you wish your 20 year old self knew uh i wish they knew that anything worth doing takes a lot of time yeah guys anything worth doing takes a lot of time coming from a prey dot io again not working on automated ai training we've got about 12 customers right now paying a minimum of five grand a month so north of 60 grand a month right now in revenue up from three grand a month just a year ago so nice growth team of four right now looking to raise maybe a million to two million bucks to continue to remove this kind of thing from human hands and drive down essentially margins for those companies drive down costs he's willing to spend up to fifteen thousand bucks to do five thousand dollar a month customer so three month payback period again as they look to scale matt thanks for taking us to the top yeah absolutely thanks for your time