#2288 She had people do AI’s work

0:00This founder had human beings do the work of AI. Humans answered questions, uploaded files, organized financial data. It helped to raise$20 million for an AI-based accounting software company. Copy her approach. Helen Hastings is the founder of Quanta, accounting software that works fast because it uses a combination of humans and AI-backed software.

0:25Helen, the interesting thing about you is that you basically had humans do AI in the beginning. How did you do it. Give me an example of something that a human being did then that today AI is doing because you noticed it back then. So we saw people just reading a lot of human written text is the biggest example of where AI has made a difference. So imagine you have purchased something with a corporate card, or maybe you need a reimbursement, you have to put a memo on that into your company's expense system. So that's just human written language. And this is something that bookkeepers were reading, and then figuring out what it meant.

1:00And LLMs are very good at that. People think that accounting is just numbers. Actually, so much of it is just understanding what's going on in the business. And so much of that is actually human written. So the example that you've given me before was you said, look, there would be a human being that would see a bill come in from Amazon and analyze it and say, this is not an Amazon purchase. It's an AWS purchase. It's still the same company, but clearly those are two different buckets of expenses. And a human being would analyze it and then do it in the beginning, right? And then eventually it was actual LLMs that were doing that kind of analysis.

1:35Yeah, exactly. So one reason that traditional bookkeeping tends to be so painful is that it's often outsourced to people that are maybe overseas and they're great, but they just don't have the context of what our software business is doing and what are the vendors that they're using. So they won't really actually understand the difference between I used Amazon to buy the coffee maker and this is my AWS bill. And that's even a company that's been around for a long time, AWS, but they can't keep up with the very new vendors as well. So when would Quantia inherit books from other providers, we see all of these things miscategorized.

2:17And it's actually very simple for us to get that understanding based on what we see as the memos and what we see as the receipt content and invoice content. Okay, so let's go back to even before you were doing this all by hand. As soon as you had the idea that you wanted to do this, you decided, I'm going to go and shadow companies. Who did you shadow? It was really a wide range of just bookkeeping firms and bookkeepers that were working with a wide variety of companies. So, like, for example, someone who did a lot of bookkeeping for, like, truckers and, like, one-off sole proprietors, as well as firms that were working with venture-backed software companies as well.

3:00So they just let you walk in and look at how they were doing their books so that you could create a software company that would do what they do? Not quite. This is very much user research phase. And, you know, a lot of these companies are they're looking for better software, too. So they're they're looking to work with with people like me who are interested in coming and saying, hey, I want to build software here to to make everyone's lives better. So it's, yeah, Silicon Valley tends to be just a very open place also. It really is. And that's one thing I really like about that is, I mean, I chat with competitors and people believe in just the more we share, the more everyone will learn and get better.

3:44You know what? That actually is a great point. I don't see this happening in the way that you experience it in many other cities where you can just go in and say, can I watch what you're doing? Can we talk as competitors? Can we have lunch? Can you introduce me to people? Okay, so you said, I'm building this, and I'm assuming also your background from Stanford to Google to NerdWallet to Affirm, I'm assuming all that background gave you the legitimacy and the credibility for them to invite you in. What are some of the things that you noticed when you were shadowing these accounting firms? And yeah, the background, I feel it was definitely helpful, and I feel very lucky that so many people were just willing to talk to me in the user research phase of starting Quanta.

4:24I was just full time going through LinkedIn, looking for finance managers, accounting managers, accountants to talk to and learn from and just ask for favors repeatedly. And it was hard at the beginning. But once I got in the groove of it, it made I really don't know how I would have started this company if I hadn't done that. So people, yes, they do see the background and that's helpful. And I think I would recommend to a lot of founders out there just just do that. just do all that user research. But it was cold messages on LinkedIn and cold emails, and I'm assuming also some friendly introductions, and that's how you got meetings.

5:05It started with warm. So it would be searching on LinkedIn, here's the title that I want to talk to, and then seeing, do I have any mutuals with this person? LinkedIn is great for that. And then reaching out to the mutuals and saying, hey, I see you're connected to X, can you please connect me? And they often would because I knew them and I'd worked with them in my career. And then when I had that call with the person, I said, who should I talk to next? Who do you recommend? And then that is how I ended up talking with different firms and people doing this work in reality. And it just, it compounds.

5:41The more people you talk to, the more you meet. And that is how I got all the learnings that I really needed to to start this company. All right, before we started recording, you told me about some of the manual things that you saw that you said, wait a minute, that software could really do this. Humans should not. What are some of those things? I think the biggest example is just how much time is spent logging into a financial tool, downloading some sort of data, massaging it, and then uploading it into a traditional accounting system. That is so much of what bookkeepers are doing today. And there's just Plaid.

6:14Sorry to interrupt. So Plaid isn't 100 % accurate. I will say that. We use a mix of Plaid that Infinicity and other tools. But the best is when companies have APIs where the data is coming in. Well, the reason that Plaid isn't enough is that there's always small things that are missing and there often isn't enough data to really understand what was that, that big transaction needs to be augmented by a receipt or some sort of understanding. And we would often inherit books or see books where yes, there was a Plaid connection, but the data was just off. And we have just built this really great understanding of all these different banks and we've been able to augment it.

6:57And yes, a lot of bookkeepers would just end up saying we have to go and log in anyway because the data coming in through the best temp automation just wasn't cutting it. I'm shocked by how many bookkeeping companies I've used over the years that have done that. They've asked me to give them access to the CSV. They have gotten CSV files themselves. It's so frustrating that that's what they're doing. Okay. And so you noticed that and you said, okay, I definitely can build software that would do it. I get now that you understood what can be done. Why did you decide that you were essentially going to build a QuickBooks or zero replacement instead of saying this exists?

7:31People have tried to unseat them for years and I'm just going to build on top of them. The biggest limitation to everything I just told you about was actually QuickBooks. So, for example, the plaid connectors that everyone uses had all the problems that I just talked about. The ability to automate things in there is just so limited. And a lot of that is because it's trying to do everything for everyone. I mean, the market penetration of QuickBooks in the U.S. is just crazy. And because it works for all small businesses, that means it's not specialized in any. And because of its limitations, there's so much manual work that is needed.

8:11So it was actually essential to me to say, we're just going to build this for the ground up or else we're never going to be able to deliver on our thesis. And then also, really, the reason I started this company is I'm a software engineer who has built a lot of ledgers. That was what I did for my entire career in fintech, building financial systems of record is the word that we use. That was my specialty. And that was what I saw as the opportunity. If we can build the financial system of record for businesses from the ground up, we can solve all of this pain that I've been seeing that has been caused by just the limitations of that existing general ledger software.

8:47Would you give me an example? Like what couldn't you do with QuickBooks that beyond obviously the Plaid connection that we just talked about? So one classic example is anytime someone changes something, all of the properties of the system are messed up. I could go in there and create something in QuickBooks that just changes the bank balance of my bank. And that's one reason the Plaid Connectors only go so far is that someone just makes some little edit thinking that they're doing something else. Suddenly my bank balance is wrong and my source of truth is wrong. So we've built something that enforces that edits are only made correctly because we're reconciling continuously all the time and something that can track changes, changes of all different types to know what is the history of what happens.

9:35So every change comes with the understanding of what happened here and only good changes sort of make it through that reconciliation process. So that's why every time I go through my monthly check in with the bookkeepers, they always will say, I'll see something and I want to go and change it. They'll say, no, no, let me do it because I might in fixing it screw everything up. Exactly. Yes. And the way QuickBooks is set up is it just makes it really easy for you to make those mistakes because it doesn't have all these checks built in. It's really just sort of a blank slate you can write into. And the only way to verify that the data is correct is people manually looking at it versus the thing that we built from the ground up is checking itself all the time.

10:19Redundancy is a big thing that we talk about. At Quantia, you need to have a bunch of redundant checks and they're running all the time to make sure that the data is accurate. For example, like what's a redundant check that catches? I guess it's the balance in a bank account, right? Right. The balance in the bank account is one. And then how are all the transactions in the bank account matching what we see in our system? So that's two pieces of data that if one is right, this one is right, this one's going to be right. However, we should do both of them just to make sure. And then also are all those date transactions matching some other system?

10:52The transactions usually in a vacuum aren't the only piece of data. They're usually linked to something else. So, for example, from my payroll system, I see that a payroll ran. And then in my bank, I see that the money moved to run that payroll. We match those things together and make sure that they're consistent. And QuickBooks does not do that. But we built that from the ground up in our system. All right. Let's go back even a little bit before all this. How did you know that this needed to be built? Where did you get the idea? So, in my past job before starting Quanta, I was a software engineer at Affirm, the fintech and lender.

11:30I was there for almost six years, joined when it was around 100 people, got to grow it through over 2 ,000 people through IPO. And really, all I did was build ledgers there. And I was building ledgers before I even really knew anything about accounting. But just the way we built things there was double entry ledgers. And it just made sense if you're running financial systems, if you're keeping track of balances and where money is and where it's owed and where it's moved and where it's moving, you build in this way, this immutable way of tracking every single change. And I saw that when you built in this way, you solved a lot of pain in just understanding the system.

12:10And then I worked on the first iteration of a firm's in-house accounting ledger. And it was piping into NetSuite, which is another, just the dominant ERP right now. But the source of truth for all of the core Affirm data was going through the in-house ledger that we built. And I saw the limitations of existing systems. And I saw just the lack of clarity that existed before our team worked on this. And I thought for a bit, maybe this is just an Affirm problem. Maybe this is just a complex fintech problem. But then when I was doing user research after I left Affirm, I thought, actually, all companies have this problem.

12:52They don't have visibility into sometimes just basic questions. How much money am I making? How much do my customers owe me? How much do I owe to others? And the root of it is the limitations with the underlying tracking of the finances. And I wanted to solve that problem because I saw if we do this, there will just be so much pain solved in just understanding what is going on in your business. The core of it is very basic. There's really just no visibility right now. If you are waiting until the end of the month to understand what your finances are, the time has already passed to make the decisions that are important.

13:32You really need to be making decisions on a day-to-day basis. And that requires understanding the fundamental numbers of your business. Yeah, those are the two things that have really made it easy for me to understand why people are working with you. The first is with most bookkeepers, it takes at least until halfway through the next month before they're done with the month. And sometimes it takes you a little scheduling time. And so you're basically about a month out from what happened. And I always wondered why they couldn't get it done faster, especially since I'd interviewed so many people whose whole thesis was software can do a better job than bookkeepers.

14:08So that's the first thing. And the second thing is when you're trying to answer a question, it is so hard to get that answer from software that you have to go to a human being who did the books and find out like little things like how much are we spending in software? Well, we got software in all these different buckets. Let me go and add that up. And that seems like the two big the two big exciting parts about this. Definitely. I mean, it's even more than a month often. So we're recording in early December right now. I mean, I talk to companies that do not have their October data yet because their outsource firm starts on their October data towards the end of November because they have just a long queue of clients.

14:49The thing is, once you're doing something manual with humans, you're not doing it every day because you just don't have enough humans to do that. So suddenly there's a long list of work. So as soon as something is manual, you are doing it on a very long time delay. So if you happen to be at the bottom of the queue, and sometimes the outsource booking firms charge you more to be at the top of the queue. So if you're at the bottom of the queue, they might not start on your data until the end of the next month and then take a while to finish it. So it even sometimes is longer than just a month. And by the time you get that information, it's just not useful at all.

15:26And we've really changed how companies see their financial data because if it's two months late, it's just useless at the speed that startups are going. You can't be operating at that pace. But once you have it available to you and close to real time, you're actually looking at it and using it to make decisions. So it's interesting how some of our customers, they used to see bookkeeping as just something I have to do. It's around tax season. But now we've changed it into something that's strategic because you have data that you need to make decisions that are critical to your business. OK, so first thing, you're discovering a problem, a pain, and you say, OK, I think software can do this, especially the way that it is today.

16:04The second thing is you're going and you're evaluating, you're shadowing people who are doing it already to see, okay, what are they doing that they may not be noticing that I think is an opportunity for us? Before you do it yourself, do you validate? Do you go to small businesses? Do you go to startups and say, if I build this, would you be willing to pay? Or do you just say, I see enough of a problem. I'm ready to go. Oh, absolutely. I recommend to any founder, ask people if they're willing to pay for something. I think it's essential to start a business. I mean, you definitely need to trust your guts as well.

16:36But understanding how customers think about things is so important for a go to market. So there's one thing to say, I know this product is going to solve pain. And it's another to understand how do people perceive it and how do you talk about it such that they are going to purchase it. OK, how much did you get? Like how much financial interest did they were they willing to pay? Was there a story of somebody who said, let me wire you now? um yeah i mean we when i was talking to people they said yeah i i would i would do this i would totally switch over uh from my existing that's it just having them say i would is enough yes yes that's a hard one because you're taking on their full books people would say sure if you can build it but you know there's always going to be some edge case some little thing that they need that they have in quickbooks like access to uh the ability to give their accountants access the ability to give their bookkeeper act, all that stuff.

17:30Well, that's one reason that the hybrid software and services model is a great way to get started on a business like ours, because that last mile edge case, we do have humans do. And it's helpful for onboarding and understanding our customers' businesses. And they also want that peace of mind of there's a human as well. So we're able to automate the vast majority of it, but they have someone to talk to who has a CPA background and works at Quanta. And knowing that we can bridge that last mile edge case with that human in the loop process, which, by the way, is such a standard way to build software in the AI era, especially in this tech enabled services industry.

18:16It just gives us the confidence that like, let's just get off the ground running, working with customers as soon as possible because we know that we can bridge that last mile. And then over time, it gets more and more and more automated. I see. And if you could just give them, I guess, Excel spreadsheets with their data, that that's enough. Is that what it was? We I don't think there's been a time where we've just handed over an Excel spreadsheet of data. The last mile things that we were doing mainly from the beginning were just looking at the data ourselves before our engine could process it and handle it.

18:51Or maybe a new financial tool that we hadn't built in integration yet or we didn't understand. Or maybe you were using the latest type of payroll system and a new payroll provider and we hadn't built the understanding of it and automated it yet. That was the thing that we would bridge manually in the meantime and have engineers do that too and have me do that. And I think that, by the way, I think that's the best way to just build a great product, just really see the dirty parts of it instead of just handing it off to someone else. But we, in terms of just doing it in spreadsheets, that's not something that we wanted to do.

19:28It's been important to us from the beginning to have it all within the Quanta platform. But sometimes we would manually get the data in there to start. But once it's in there, it's in this well understood system that we've built that unlocks understanding for our customers. You know, when I interviewed the founder of Zenni, that's an odd AI bookkeeping company. This was back in April 2021. And I thought it's so compelling. But at the same time, I had a hesitation. I said, I don't know that I want to trust your software. And what the founder said to me was, you know what? Worst case, you still have everything in QuickBooks.

20:02So you can always switch this out. And I said, OK, that's a reassuring thing. That's what I meant with you. Was there any kind of backstop there in case people didn't like what you were doing? Sure. So by the way, we get this, the trust problem all the time. It's so enormous in accounting and we still get it every single day. We actually recently announced our series A and announced a new product, which is agentic reporting. So it can really answer any question about your business. And I got so many messages that day of this is fantastic. Does it work? and if this works, I will buy it. But they don't trust that it works yet.

20:39It's almost too good to be true. So this is a big problem in the AI era, I think, is these too good to be true problems. But when it comes to the QuickBooks and bookkeeping, actually we do have a product that we call a QuickBooks backup where if you, for whatever reason, want to do an export into QuickBooks or even keep a QuickBooks up to date, we support that. And it's a good way to get people in the door, especially we work with a lot of companies who have had years of previous bookkeeping history. And then they come to Quanta and they're kind of afraid to give up their QuickBooks. And I think they should be.

21:18You think of companies like Bench going out of business and people have no access to their history of the most critical business data. So I definitely understand that there's that hesitancy. So one strategy that's worked well for us is saying, you know, we'll keep your QuickBooks up to date for you. Quanta becomes the source of truth. You can't touch the QuickBooks anymore because that would mess with all the properties of our system that I was discussing earlier. But we'll let you keep that up to date, keep that live with Quanta data as if someone was keeping it pristine from the beginning. that then we find that our customers will often just shut it down after they realize, okay, I don't need this anymore.

21:56Why should I pay for this system on the side? So funny. Like you mentioned Bench. I'm pretty sure we're a sponsor of mine because bookkeeping is so popular. So they sponsored. Then all these companies I'd worked with in one way or the other. Pilot, I think, was a sponsor. And I really loved what they were doing by adding automations into QuickBooks. I invested in Indonero. I think it was the first startup I ever invested in. They were going to replace QuickBooks, which at the time and still today, I can't stand. I hate QuickBooks. But then there were all these issues with it. They eventually all seemed to come back into QuickBooks.

22:31And the one company that survived is I interviewed the founder of Zero years ago. And for some reason, they're still around. I don't know what the difference is. Why are they around? And then I'll ask you about getting into the space where everyone's dying. Why do you think Zero survived? So Xero actually has a great market penetration outside of the U.S. So they were founded in New Zealand, actually, and their international support is really incredible. So they're very popular in other countries. QuickBooks is the dominant player by far in the U.S., but Xero has a great international footing.

23:07So I'll talk to some U.S. companies that use Xero just because they started in another country or the person managing the books was from another country. And so what do they do differently? They work for other countries. um really i think so much of it is about just knowledge right mind share people know about it because it is just a dominant in that space and uh every country has a little bit of difference in how do you set up what uh what categories the tax authority in that uh country really cares about because a lot of the early stage bookkeeping is it's an input into tax and because if your tax authorities different as it is across different countries, then maybe you want to structure things differently.

23:52Okay. Yeah. They also, I think, were cloud before QuickBooks was fully cloud. They also did the thing. QuickBooks is still not even fully cloud, which is very funny. So a lot of people still use QuickBooks desktop. That's still available. Yes. I think they might've started deprecating it very recently, and it was a huge deal. A lot of People were up in arms about, oh, no, QuickBooks is saying that they're going to start not investing more in QuickBooks desktop. Because people do assert that QuickBooks desktop can do more than QuickBooks online. It's funny, the acronym that everyone refers to for QuickBooks is QBO.

24:32Yes. The O stands for online. Yes. And that just shows how dated of a product this is, is that the word that it's online is such an important characteristic of the company. Right. But I talked to so many accountants and bookkeepers who are just love QuickBooks desktop and assert that it is better than QBO QuickBooks online. And again, I think it just it shows how dated this industry is. I'm going to be one of those people. I actually think if you're doing it individually and you're not trying to collaborate with other people, their desktop has less of the garbage that all of their online stuff has.

25:08And I understand why people would want to get rid of all that garbage. Every time you log in, there's more cruft, more of them trying to analyze how you're doing, do a Qantas. Is that what it is? No, Qanalytics or whatever it is that they're using to figure out if I'm happy with them or not. And I keep telling them I'm not happy with them and they haven't changed anything. So why do you keep popping up that rectangle on the bottom of the screen? I mean, all right, I get it. But aren't you scared now considering how bad they are? They've been so bad for so long and still pilot, Zenni, zero to some degree.