Ep. 191 - The Next SaaS Wave: Don’t Sell the Tool, Own the Outcome

Ken Lempit: Welcome to SaaS Backwards, a podcast that helps SaaS and AI CEOs and go-to-market leaders, accelerate growth and enhance profitability. Our guest today is Sai Dhanak, CEO, and co-founder of Deduction, an AI powered tax advice and filing platform. The combines intelligent agents with licensed CPAs to deliver proactive, always on tax services at a fraction of the cost of traditional firms. Welcome to the podcast Sai.

Sai Dhanak: Thank you for having me. Ken Lempit: Yeah, really excited about your company and this episode. But before we dig in, I'd really love you to tell us a little bit more about yourself, your entrepreneurial journey, and your company Deduction. Sai Dhanak: Yeah, happy to.

My name is Sai. Currently the co-founder and CEO of Deduction. As you can probably tell from my voice, I'm originally from London, England. Grew up there and started my first company there called Caribou, which was acquired by the toy company Mattel.

And after that it got me a one-way ticket and visa to America, which was very fortunate where I helped my friend, start a company called Latch in New York, which is a keyless entry company for apartment buildings. In other words, allowing you to unlock your apartment door with your phone and, pay rent and things like that. We took that company public in 2021. I was the chief product officer.

The whole whole company was about 500 people and really a journey of the lifetime. And I left in 2023. Again, I was super fortunate to leave. And be able to start something new.

Took me a couple years to, to really build conviction on what I was gonna do next. And ultimately my co-founder, who was actually with me at Latch Jonathan Kieliszak, whose dad was a CPA, his whole life triggered a kind of obsession with the fairly outdated accountant industry that ultimately is what 55% of Americans choose when it comes to tax advice and filing. In other words, most people in America want an accountant. And what is happening over the next 5 to 10 years is that most accountants are retiring.

And so we're facing this crazy supply demand disparity. And we realized this was a once in a generation opportunity. To build a different kind of tax company for consumers that would solve three problems. One you wouldn't have to wait weeks for a response from your accountant.

I think a few people Ken who are listening, that probably resonates with them. Number 2, you don't have to pay exorbitant amounts of money. Upwards of a thousand dollars shouldn't have to be the case. And number 3.

Have an actual accountant that's proactive. And so deduction.com solves those three problems. Think of us as like an agentic, H&R Block, or a Claud for tax.

The key thing is that we have real CPAs and tax experts in the loop to kind of review high impact decisions and findings. And so, that's Deduction in a nutshell. Ken Lempit: It's really exciting and I have to say, I'm always amazed, actually, how people find these greenfield opportunities, even in a really traditional space like tax prep for consumers. So it's a really exciting opportunity.

I don't think there's much like what you're doing out there, and I'm really excited to dig in with you. So you said that, many of us have experienced this sort of slow and expensive and reactive accounting, but that also was your own experience, right? So, what made you think that, or believe, as you said, conviction, you had conviction. So what made you believe that AI could actually replace this traditional service, not just be sort of improvement around the edges?

Sai Dhanak: Well, I'll tell you two things. One I'll tell you. We add the theoretical ideological. Instinct, we use the AI tools.

We, we are using ChatGPT. That would be the academic answer to your question. But I would say that the subjective answer to your question is, we prototyped something very simple and what we prototyped was, the very, very specific instance in how people refer an accountant to each other. So the, the, the way that people get an accountant is they ask friends and family, and then the friends and family will then email introduce you to their accountant, and our whole vision would be the only way you could change a very service labor intensive market that's very relationship based is if the AI is almost indistinguishable from a human experience.

And so what we prototyped was very simple 'cause it was only two, two of us. And it was like one week or two weeks in with the idea. And what we prototyped was that we gave our AI agent an email address taylor@deduction.com, which by the way hasn't changed since.

And we prototyped this ability where I could email Intro Taylor to anyone of my friends, and Taylor would react to that and get going. And the magical moment happened where I emailed Introed a friend of mine just to see what happens. And the agent Taylor BCC'd me and responded, saying, Thanks, Sai, for the intro, Jonathan would love to talk about your taxes. Please sign up here.

And at that moment, I felt like it was so at the time, felt so magical because an AI was BCCing someone and acting it's such a human thing. And at that point we said, okay, it's definitely gonna, it's not gonna be easy to do this. But the key thing was that, is there magic in what we're doing? And that felt magical.

As funny as that sounds, it felt pretty magical. Ken Lempit: Well, no, I mean, it's like. Your, your idea was manifested right by this simple email exchange. And I, I think, if I draw from our experience trying to make emails generate response like in marketing or sales sequences, the simpler those emails, it seems, the better.

So here you have how, actually a human might write an email, right? Just a handful of words, but in the parlance of a normal human conversation. That's really cool. And I, I did see by the way that you have Taylor, I think that's, it's a great persona.

Sai Dhanak: My wife, I actually had a different working name, but like any great startup our spouses have undervalued contributions. And I have to say here officially that my wife did come up with it, and she deserves the credit, as she always says. Ken Lempit: In our prep. We talked about how most SaaS companies or you know, these new AI companies, they're building tools for others to use, and in this case, building tools for accountants to help them in their practice.

But deduction here is actually the tax firm, and I'm wondering how that shift from SaaS, to a SaaS, an AI delivered service change how you build and go to market. Sai Dhanak: You know, I think the whole SaaS apocalypse is obviously overstated. At the same time, I think people are realizing that the speed at which you can build these tools means that the traditional SaaS moats are eroding a little bit. but I think the third thing is that there's some things that have always been true, which is that I think a lot of people think B2B SaaS is easy.

You just all, you build the thing and then you go, you go door to door and you sell people and voila, you have massive ACVs and low churn. And the reality is like, B2B is really hard and I think B2C gets a bad rep because a lot of people lost money on social media apps. But the way we approached it was kind of first principles, which is, what's a good business in this market? And sometimes B2B is a fantastic business in a specific market.

We've seen incredible B2B companies, recently Cursor, Figma, others. But in this market we had conviction that in a market where the existing professionals are retiring, you probably don't want to sell. Software to a shrinking TAM. No one's ever been like a shrinking TAM?

Sign me up. Ken Lempit: Let's sell to that ideal, Sai Dhanak: Yeah. So, I just wanted to start with like, my, my reasoning as to why we decided to do something that on the face of it sounds much harder, which is you not only, you are the consumers of your own SaaS basically. Right?

And you have to build the SaaS product and you have tosell it internally, let's say, and you have to go get the end customer right, so you the full stack play. Now, as I say, there's good reason why I actually think this is easier to do in many markets today and more enduring as a business than just SaaS to a let's say a professional, right? And we can dig into that in a bit, but, the way we sell, that's different. And the way that our tooling is different is, on the one hand we have a team that's dedicated to demand generation.

so we have to get our own customers, right? So that's how we are different in one sense. I will say that one of the ways we shortcut this is that we actually acquire the books of business from retiring CPAs, which is a very sort of an accelerant to what we want to do. And the second thing is in terms of kind of product and tooling, as I say, we are kind of the, what's really amazing, and I think a lot of B2B SaaS founders will resonate.

By the way, I was B2B 2 C, right at Latch. We sold to property managers software. Is , when the subject matter, experts actually work for you. The pace at which you can develop your product is exponentially higher.

So instead of, let's say we develop a feature, we then have to arrange a meeting with our customer. We then like go through it. The customer gives us some info. Then we go to a different customer.

A customer gives us different information. So it's the long lead times of iteration. I'll tell you right now, every afternoon we have engineering and our CPA team meet share learnings, and we are developing features literally on the order of two to three days. So we've already built today a full workflow management system for CPAs and AI.

We've built integrations with various different tax software. We built our AI agent. The speed is just phenomenal. So I would say the number one thing is when you go fully vertical, you cannot underestimate the speed at which you can move relative to having your customer be an external professional.

Ken Lempit: the vertical SaaS, is well known as a kind of better route to market, more profitable route to market. The multiples are higher, certainly in B2B. Here, you've taken this service that consumers have a demonstrated willingness to pay for. Also, I think a lot of the failures, you alluded to some of the social apps, there might not have been a willingness to pay, whereas here, people are willing to pay they overpay by some estimations, So, here's an opportunity totake people on a journey they're already kind of familiar with, but with a new way of getting it done.

So I think that's part of the secret sauce here. One of the things I wondered, is there a possible future here where you might offer this as a back office solution to other CPAs, or are you only gone for, market share in the prep work itself? Sai Dhanak: Never say never. I think we have some interesting ways that we would want to do that.

I don't think we traditionally would want to. On the face of it, just go and white label for any CPA firm. That said, I think there are different ways to work with the industry, number one and number two is I think there's a pretty long tail of let's say adjacent partners that we would want to work with. So not just accountants, but it could be wealth advisory platforms, banking platforms, HR platforms, mortgage lender providers.

The long story short here is, every single company, it's a bit like when smartphones became a thing and every company realized that they needed a mobile version of their website and they probably needed like some kind of app, right? And so now you can see like every company, every major FinTech company, even like the old school mortgage lenders have an app, right? I think it's the same way with AI agents. I think every, every company eventually is gonna have an AI agent.

Even your old school mortgage lender or your old school insurance provider, they're gonna have some, it may not be the, the best working agent, but they're gonna have an agent. And my conviction is that we will probably be the accountant agent on record for those other agents. And so we're gonna have a, we're gonna live in a world where all these agents are just talking to each other to satisfy different needs for end customers. And, that's how we dream of working with, as you say, the back office, we would like to be the chosen back office accountant agent for consumer for let's say consumer FinTech platforms.

Ken Lempit: Neat idea. So in our prep we talked about professional services firms scaling and being profitable based on labor arbitrage, right? So it's a markup on labor and also finding less expensive forms of labor. So you talk about AI arbitrage, and I'm wondering how that works in your business model and how does it reshape pricing?

I've talked about the death of the billable hour, and I think ultimately we're moving to a world where these industries that are relied on billing hourly are gonna start billing with by outcome. And so, we priced from the get go to be based on outcome. And so we offer one fixed price, $499 a year. Sai Dhanak: We'll likely have additional tiers in the future, cheaper and more expensive.

And with that, you get unlimited advice. So you won't get billed per hour for advice, and you get your annual filing and you get proactive strategies and planning throughout the year. So immediately, that's our immediate strategy to kind of flip things on their head. And that's why I say it's more than labor arbitrage.

'cause in labor arbitrage, you may be, you're, you're trying to get cheaper labor so that your hourly cost is smaller, but your hourly price is the same, if not more. For our opinion, AI arbitrage is where. Yeah, arbitrage may not even be the right term, potentially but it's where you basically flip the switch to outcome-based pricing. And so then your challenge is actually much higher.

It's not just about cheaper labor, it's, can you effectively have software be powering 80% of the business while you reserve amazing subject matter expert talent for, the sort of review and occasional, stepping in, if that makes sense. So, it's sort of a fundamental adjustment of the entire way that the organization is structured. And that is why, there's a good article that came out by the CEO of heavier, I think it was yesterday or the day before, which was that, the, steam didn't immediately make these mills ultra productive.

They had to redesign these mills to really take advantage of steam power technology. And so in other words just adding a bit of AI into a company doesn't make it, tremendously efficient. You actually need to redesign the entire institution itself. And so that's what I would call AI arbitrage.

Ken Lempit: So I'd like to maybe make it a little more metaphorical. So we have these AI agents that are sort of our junior tax preparers, Our tax prep experts. But they're maybe not. Fully the CPA with 10, 20, 30 years of experience and all the context and history that can add to the work of prepping returns.

I guess the expectation or the promises that you're gonna get a better outcome as the user, right? You'll have a more accurate, more encyclopedic view of the tax regulation and contemporary view, and as well, great expertise on top of that. Sai Dhanak: Yeah. Yeah.

Or another way to put it is just giving accountants, Claude and chatGPT will not fundamentally change the pricing, the proactiveness and the responsiveness of a service it'll make the company more profitable. It may make them slightly more responsive, but if you wanna fundamentally change the entire experience, you need the AI to become the professional and have the professional become more of a manager. Ken Lempit: Yeah, that's a great way to say it. I wanna just circle back on one thing.

How did you go about designing this mix of AI and human oversight? How do you make that happen? Sai Dhanak: it's actually probably the most exciting, one of the most exciting things that we built. And I would put full credit to our, two founding engineers, Emily and Gregor.

But, from the get go, we realized we needed a seamless way for the agent to escalate issues to a human in the loop. And so what they developed was what we call what basically a internal only platform where our tax team can monitor the AI agent and the agent can surface issues pretty much as tasks. So the way to think about it is we basically have an internal built Trello board, and again, this is back to my point about how fast it is to build SaaS products now. We've built effectively an email based customer support platform similar to Zendesk.

We've also added to IT workflow management. That's a bit like Trello. And on top of that, we've added billing, e-signature, and various other integrations. We've done that all in six months.

We've basically built like a super SaaS product that if I showed it to any CPAs, they'd be like, I would love to have this right now. And so we built this internal platform to enable this sort of AI human collaboration. The one thing I'll say to say is that even though SaaS, and let's just say software's easier to build. I can't tell you how important, that human taste still remains so important when you design these interfaces and these products, even if it's for your own internal team, which is, we built it for our own tax team.

And how much thought and work still goes into that despite the fact that Claude will tell you five, six different ways to design the product. So that's effectively how we have humans and, and AI work together. And then the last thing I'll just say is from the beginning, our conviction is that our agent must feel like an accountant, we have to meet consumers where they are, otherwise, they're not gonna believe that this is an intelligent AI and so our agent is fully email based.

And so effectively we've built an email system where the AI can send out emails back and forth. While also having humans review emails that let's say are high impact or, need review. Ken Lempit: That's very cool. I wanna move to this thing you mentioned about acquiring books of business from retiring CPAs.

I mean, that's amazing. But I'm, I'm wondering how has that really been effective in this B2B2C kind of mindset? And how about bringing those customers along with you on that journey? What's like the onboarding been like?

Sai Dhanak: you know, I would say it's been a very efficient way to build the business. Mostly because I think this is one of the rare industries where you could just buy a book of business, meaning you can just buy clients and they transfer over to you without any assets, liabilities, systems, processes. Talent that maybe you don't need. So, it's a rare world where basically you can think of it like a giant referral where you basically pay a CPA or a tax company and say, you are retiring, I'll pay you some amount of money over some amount of years, and in return, you're basically going to persuade and, and communicate to all your existing clients that they should move over to you.

And of course, with that comes all their data, right? Which is extremely powerful. It makes things easier. They don't have to go find a new CPA and tell them all about their life.

We already kind of have all that information. So it's a very, very efficient way. My perspective is. If you are gonna go down the agentic services route, so you decided I'm not gonna sell SaaS to the service providers, I'm gonna become the service provider myself, then you know, I think there's an immediate question as to, whatever industry you're in, does it make sense?

How are you gonna go zero to one. Are you gonna go acquire customers slowly? Are you gonna try and rapidly build a base of customers? And I would say acquiring some of these firms, whether it's in accounting, whether it's in law, whether it's, I don't know, landscaping, whatever it is, I highly recommend it because You know, they sell at very good multiples and you kind of immediately bootstrap and accelerate your progress.

So that's number one. Number two, on your question about onboarding, again, because we've been thoughtful about how our agent manifests itself, the fact is that people are pretty happy to email our agent. And we don't require you to download an app. We don't require you to do kind of all these different, jumping through different hoops.

All we say is, Hey, this is your new tax team. They got 20 years experience. Here's the new email address that you're gonna use to communicate with them. We do use an AI virtual assistant to make things better, faster, cheaper, and, of course some people have their concerns and we've done a lot of work to develop the right materials to acquiesce those concerns.

But the vast majority of people want a good experience. And my opinion is there's no way you can provide a good experience in a market where labor is basically becoming non-existent unless you have a very good agentic platform. Ken Lempit: That's a really important insight that you have to have you know, the great customer experience. And also, just going back to the increasing scarcity, right.

Of those professionals. wanna land the episode on just one last question, you have this always on proactive tax advisor. Is there any limit to what you can do in this space with an agentic CPA? Could you also go perhaps to business returns?

I mean, does it have to be limited to consumers? I mean, is there more greenfield within your vertical market? Sai Dhanak: The more accessible and better and cheaper a service or product becomes the more demand there actually is for it. And my perspective is that there's probably more opportunity that we don't even know of with the consumer market if we can put an accountant in every American's pocket.

What I mean by that is. What is the ideal accountant? Is gonna be the question, and is it just filing your taxes? I don't think so.

I think the ideal accountant is gonna do way more for you and will probably work with other AI agents to deliver services to you. Imagine an AI agent or that knows everything financially about you in a secure, private way. Is able to help you with additional situations that a normal accountant just doesn't have time for. So, to give you an example, if you're thinking about buying a house and you're shopping around mortgages, can a great accountant help you with that?

If you're, if you're thinking about buying a car, could a great accountant help you with that? And if you're thinking about setting up, a trust or a state, can a great accountant help you with that? So, I think that's what we're really interested in. I do think that there is a blurry line between individuals and businesses.

Someone that owns an escort, a solopreneur. Are they a business? Are they an individual? So we care a lot about those people as well.

I think where we probably today draw the line is, multi-member S corps. 'cause we think that that just opens up a whole different world of bookkeeping, payroll. And I think there's a lot of great companies and services that already do that. So I'm interested in the, unknown unknowns of really doubling down on the consumer.

But who knows? I think running a company you've gotta be agile and I may come back next year and say, you know what? We now have an SMB tier. So, that's kind of where we're at right now.

Ken Lempit: So I think we'll have to stay tuned and hear more from you and Deduction. So if people wanna learn more about Deduction, how can they go about doing that? Sai Dhanak: deduction.com or if you feel even easier, you can email taylor@deduction.

com. Ken Lempit: That's awesome. Thanks so much. Really exciting episode.

Can't wait to see how things evolve for you. If people wanna reach me, I'm on linkedin/in/kenlempit and my advertising and advisory firm for SaaS and AI software companies is Austin Lawrence. We're at austinlawrence.com.

And if Sai hasn't convinced you to subscribe to the SaaS Backwards podcast, maybe no one can. Sai Danek, thank you so much for being here today on SaaS Backwards. Sai Dhanak: My pleasure.