Ep. 194 - The SaaS AI Trap: Fast Answers, Bad Decisions

Guest: KG Charles-Harris, Founder & CEO of Quarrio  

--  

The SaaS AI trap is believing fast answers are good enough when the real advantage comes from trustworthy, decision-grade intelligence.

In this episode of SaaS Backwards, Ken Lempit talks with KG Charles-Harris, founder and CEO of Quarrio, about why most AI tools fall short in enterprise environments where decisions need to be accurate, auditable, and actionable. KG explains the difference between probabilistic AI and deterministic AI, and why that distinction matters far more than most SaaS leaders realize.

They also explore why business users do not want more dashboards or more software to learn. They want answers to questions, delivered instantly, in a way they can trust. The conversation covers Quarrio’s long path to market, how enterprise trust is built through founder-led sales, and why compressing the cycle from data to decision to action may become one of the biggest competitive advantages in SaaS.

Key takeaways:

• Most enterprise AI tools are fast, but not reliable enough for decision-making
• Deterministic AI is better suited for auditable, enterprise-grade answers
• SaaS users want answers, not more dashboards or reporting delays
• Decision velocity may become a major competitive advantage
• Founder-led sales and trust are critical in early enterprise go-to-market

–

Growth stuck? Get a free SaaS GTM Checkup

---

Not Getting Enough Demos? 

Your messaging could be turning buyers away before you even get a chance to pitch.

🔗 Get a Free Messaging & Conversion Review

We’ll analyze your website and content through the eyes of your buyers to uncover what’s stopping them from booking a demo. Then, we’ll give you a personalized report with practical recommendations to help you turn more visitors into sales conversations.

And the best part?

💡 It’s completely free.

No commitments, no pressure—just actionable advice to help you book more demos.

Your next demo is just a click away—claim your free review now.

Transcript

Welcome and Guest Intro

SPEAKER_02 0:06

Welcome to the Stats Backwards Podcast, where we reverse engineer the success of fast-growing stats firms and explore strategies DMOs and CEOs are using to drive their businesses forward.

SPEAKER_04 0:24

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 KG Charles Harris, founder and CEO of Quario, an AI lab-turned enterprise software company building a deterministic AI platform that enables users to query complex data, generate auditable insights, and make faster, more reliable decisions.

Quarrio Mission and Data Access

SPEAKER_00 0:54

Welcome to the podcast, KG. Thank you very much, Ken. It's an honor to be here.

SPEAKER_04 0:59

Yeah, I'm excited to dig in with you. This is a really going to be a great episode. But before we do that, could you please tell our listeners a little bit more about yourself and your company, Aquario?

SPEAKER_00 1:10

So I'm a multiple-time founder and have never been smart enough to do the same thing twice so that I can learn from my lessons. This time, we're building a deterministic engine to interact with structured data sources. And the reason we're doing that is because an organization is really a decision and action machine where a group of people have come together to make decisions and take actions that enhance the group. And as such, we need information to make those decisions. We've spent billions, hundreds of billions actually, centralizing data and data platforms, but we just can't get access to that data to turn it into information when and how we need it. And that's the point of this company.

BI Pain and Customer Research

SPEAKER_04 2:01

Well, uh, you know, uh it's an interesting description about what a business really is and the nature of managing it. And I really agree with you. You said this is sort of an entirely new thing, but you did come out of the business intelligence world previously. I did. And I guess your experience was that getting answers from data was slow and painful. What was the core problem that you saw that existing tools couldn't solve?

SPEAKER_00 2:25

If you're familiar with the company Domo, my last company was very similar to Domo, and we sold it when Domo went out and raised a huge series A for the time. And we said, Well, a price has been established in the market. Do we want to go out and raise money or do we want to sell now that a price has been established? We chose to sell. The essence of the problem that we're facing was that I, as the CEO of a business intelligence and data integration company who built the software, had the team who built the software. I mean, no one in the world is better than we are at it. I still cannot get access to my information when and how I need it. And that was kind of an epiphany after we saw the company that there must be a better way of doing this. And then we started working on research and going out to our past customers. We ended up, my past VP of marketing and I were ended up interviewing about 120 of our past customers, their executive teams, their business unit leaders, analysts, supply chain, IT teams, sales teams, etc. And what we realized was business users just want an answer to our question. They don't want to learn how to use different tools, they just want an answer now that influences the next step of what I'm about to do now. So that was the first major lesson we got. The second lesson we learned was, and this is in the days pre-TikTok, is that if any piece of software takes more than three minutes to learn how to use, the user adoption falls precipitously. That was pre-TikTok. We assume that three minutes is long nowadays, and so we need to create something that is so intuitive that people on all levels of an organization can access information at will using what they already know. And that was the challenge we went towards. And then doing that, since we're going to make decisions on this information that we're retrieving, it has to be accurate.

Founding Story and Technical Roots

SPEAKER_04 4:49

So that's really fundamental insight. I think it's it's interesting in that not every founder, not every product manager thinks about how important simplification is, right? How important it is to make software completely easy to use without the user manual, right? Software no longer comes with the user manual, and at least good software. That's a really powerful insight. And I'm wondering in your prior companies, did you face that same kind of adoption hurdle? Did you had you previously addressed it? Or was this new for Coreio?

SPEAKER_00 5:25

So the first data company I worked in, I co-founded with a professor of genomics, a company dealing with gene sequencing in the agricultural supply chain and for animal breeding, specifically fish farming. This was in Norway. And that's data. Genes are data. Genes are actually information. And these genes that are encoded in us actually has a lot of knowledge about how we should relate to each other and relate to our environment. So that information is encoded in data that's encoded in genes. So that started my thinking about the fact that everything is actually data and information. And for us to actually interact with each other, that's what we need to do. And then I happened to meet a gentleman called Jim Cates, who is our chief product officer at the company. And actually, he led the team that created SQL or standard query language, Tevolid DB2, IBM Watson, built on the line technologies for IBM Watson.

SPEAKER_03 6:38

I think early in my career when I was at IBM, I heard that name.

SPEAKER_00 6:41

Yeah. He's a luminary. He then went from the research and development side over to the implementation side and became the CIO for companies like Silicon Graphics and Synopsis, which designs the chips for Nvidia and so on. And he looked at both sides of this the data to information to knowledge continuum, to the ability for us to actually extract that and utilize it instantly in our organizations. And we decided to join forces on Quario is the result of that.

Deterministic vs Probabilistic AI

SPEAKER_04 7:18

So, I mean, this is a really interesting founding story, and it's a little buried in the description. There's a huge amount of user research that went on here, right? So and we've heard that in there's almost 200 episodes of the podcast now. So really understanding your customers, who they might be and what's really important to them, and then having a really powerful technical foundation and technical lead, even if that's not you as one of the founders, having someone on the team who can truly lead from a technological standpoint, those two ingredients are really important. I want to turn our attention to kind of where we are today, because there's a lot of tools out there that are being pitched as AI native, AI first, and AI to help you do things, and also LLM-based tools to query data, right? Or to give us content or build agents around to do work for us. So I want to know what Quario does, especially for enterprise use cases, that some of these other tools really can't do. And I know that the part of the embedded deterministic answer is where you're going to head with that. But I think you need to help us understand what's unique about Quario versus the tools that are getting so much of the attention these days.

Decision Grade Intelligence and Security

SPEAKER_00 8:38

Let me take a step back and talk a little bit about AI in general. AI has been around for a long time. Most people don't know this, but it it actually started with in in the 1950s. And based upon that foundation in the 1950s, two differing approaches to AI was developed. One, which was called a kind of a rule-based type of approach built on symbolic systems approaches. And the other one was what's called the probabilistic approaches, which started with machine learning, then neural networks, then transformer-based technologies on which LLMs are based. So these two different aspects of AI have different characteristics. And what we must be careful about doing is avoiding putting square pegs in round holes. When we're talking about probabilistic AI like LLMs or SLMs and SLMs with RAG, for example, or retrieval augmented generation, you can approach relatively high levels of accuracy on that. But in reality, the top models today deliver between 65 and 85% accuracy. If you're making decisions or automating processes in your organization, that's introducing risk to an already risky proposition, which is running a business. So we have to ensure that no risk is added by adding these types of tools. And the only way to do that is to be able to have auditable compute and transactions so that I can know what happens and automate the identification if something goes wrong. And that's the essence of having a deterministic system instead of a probabilistic system, because multiplying a probability with another probability and doing that 1.7 billion times doesn't make it more accurate. It just makes it more difficult for you to identify when it's inaccurate and how inaccurate it is. And that's why all the trails and so on is very difficult for these kinds of tools. So as we are now thinking about things and thinking about enterprise, what is it that we need in the modern world where we need shorter lead times? So that they can actually make a decision and take an action then, versus waiting two weeks for a report that is going to then precipitate another report. So how long can we wait to make decisions and take actions in enterprise?

SPEAKER_04 11:35

In our prep session, you called this decision grade intelligence, which I think is a really interesting idea, sort of a marketable idea. There's a lot of availability of these tools, right? We can query almost anything with these tools, and now we can build connections to our core systems, whether it's ERP or CRM, right? So we can now not only query the wider web or get information based on what's in the wider web or news, but also our own data manufacturing within the enterprise. But what I hear you saying is that not all these tools actually give us what you call decision grade intelligence. I think that's something I'd like you to comment on, you know, how you feel your company is doing that and maybe initial use cases that you want to bring to market.

First Use Case Sales and CRM

SPEAKER_00 12:25

I think that there's a lot of value to what's going on in the LLM and SLM space. And I use these tools every day myself. So they're extremely valuable if I want to brainstorm, if I want to write something, if I am just doing some research or something like that. But when I'm making a decision about an action I want to take or I want my team to take, I want that to be based on accurate information. How do I guarantee that that's the case? Or if I want to automate a business process or a workflow, I want that to be accurate. Think automating a payment process and having hallucinations in that. So that's unacceptable, as we know. So my point in all of this is that we ensure that as we're moving forward, that organizations actually have that decision-making ability and that they can feel secure in that information is right. And just to add to that, security is twofold. Whenever I use an LLM, I need to be conscious of the fact that that LLM is listening to everything I'm doing. Most businesses' core intellectual property is in their business processes. How does Goldman Sachs do things versus Morgan Stanley, for example? These businesses on the surface look similar, but how the businesses actually function internally is what is their secret source as a business. If I'm using an LLM and I am now telling that LLM what my secret source is, my business is now more at risk. So what we have done is that we have created a situation where we don't take the data and listen to what you're doing. We just enable you to get information from that data to monitor your business process across your organization. So there is no security risk introduced by cybersecurity because we sit behind the footprint of your organization and there's no data leak out into the outside world from your organization. So what does that do? And how do we use this? Well, the first thing we're doing is that we're launching this in against CRM type use cases in the enterprise. We happen to be have been incubated by Salesforce for several years, and so it's a natural place for us to go because the sales team needs information now. They are more than anyone else, the people who cannot afford to wait in an enterprise. So we need to enable them and give them what they need so that they can create call lists, that they can, while on the phone with clients, look up things that is inherent in those data sources, whether support cases or competitive information or anything like that. They need to have that ability to do that in near real time. And this is the first use case that we're addressing in quarry.

SPEAKER_04 15:36

Also, I think it's a great place to start because the sales organization is the engine for these businesses. They have the ability to invest in the effectiveness and efficiency of the sales team, right? So it's a great place for software companies to do their business.

Sponsor Break Go to Market Checkup

SPEAKER_00 15:52

Well, we actually had a long discussion about that, Ken, internally, where some people wanted to go towards finance, for example, because finance would really value our deterministic capabilities. My position in these discussions was you make sales happy, you make everyone happy. But if you make finance happy, you've made finance happy. And it stays there. And so that's one of the reasons we're starting with sales, because sales is the driver, as you said.

When to Go to Market After R&D

SPEAKER_01 16:23

If you're building a SaaS company, here's some data that's certainly worth paying attention to. According to Kyle Poyar's research across 6,500 software companies, only about one in five ever reach 5 million in ARR, and just one in 10 make it to 10 million. Now, those are some pretty sobering numbers. If you've got funding in a solid product, but you're still missing revenue targets, the culprit is almost always somewhere in your go-to-market. Now, maybe you're losing too many deals to no decision, and many times pricing hasn't changed and it's opened the door to competitors. And often sales and marketing are hitting their activity KPIs, but that's where the good news ends. Now, these are all solvable problems, but you have to know where to look. And that's exactly why we built the SAS Doctor's Go-to-Market Checkup. It's a free diagnostic where we assess 12 critical components of your growth engine, from positioning and pricing to your sales tech and metrics. We'll come back to you with a clear picture of what's holding you back and what to prioritize next. No 80-page decks that you'll never implement, just a sharp, actionable read on why you're stuck and what needs to change. So if your product should be growing faster than it is, check out the link in the show notes and let's talk. And now back to the podcast.

SPEAKER_04 17:47

So you've been working a long time on building this company, right? Correct. This has taken the better part of a decade. Correct. In RD and building the capability. So how did you know you were ready to go to market? I think this is an interesting question in a lot of businesses. It's like, do you have enough to be able to sell it? Or do you need one more capability, one more breakthrough? How did you make that decision?

Founder Led Sales and Trust

SPEAKER_00 18:15

You know, that's more of an art than a science, Ken. We were working with very complex technology, and we had to have a number of breakthroughs, and therefore our intellectual property portfolio is pretty strong. But you know, there's always this thing: just one more thing, just one more thing. And really and truly, for us, we have done a number of POCs and pilots and so on throughout the years with major companies, auto-parts manufacturers out of Germany, oil companies out of the Middle East, major utilities, investment banks. And so we have done all of this to really test the system in various environments globally. But there's always stuff to do. You know, it's like a child. When is the child an adult? What's that point? That is the question. And so at some point, you just need to make a break. And as long as clients like it, then it's time to sell.

SPEAKER_04 19:13

I want to keep on that go-to-market thread just for a little bit. You've been started with founder-led sales and you know, going deep into your network, you know, the founding team's network. Tell us what's working in those first enterprise conversations and what are the steps that founders need to take to get sales going on the outset?

GTM Channels Partners OEM SDK

SPEAKER_00 19:35

So, in my mind, for an early stage company in the enterprise space, which differs very much from the consumer space, where the enterprise space is more about trust. And if you don't have trust in the enterprise space and can show that you're trustworthy, how do you sell to enterprise? So you both have to satisfy a need as well as show that you're trustworthy. And the best way to show that you're trustworthy when you have no customers or any costing attraction is to a warm introduction from someone who the target already is trusty. So that has been our advantage, really, that we have some senior people in our network that can make introductions to people who have been patient while we have been toddlers, and so now we are growing up and becoming teenagers and maybe adults. So that's really their situation. And I think that the trust situation is so underplayed and will compensate lack of trust with marketing dollars.

SPEAKER_04 20:49

Yeah, you you've got great relationships, you have trust built up with people that can influence the outcome. But you know, where are you focusing first? Direct sales through partnerships or OEM deals? You know, where is the revenue gonna start to come from?

SPEAKER_00 21:04

So we have a little bit of a problem in that we are a very broad technology platform. And so we can be used for thousands of use cases and in thousands of types of situations. And so we have a multi-pronged approach where we've initially sold some direct clients. Now we're beginning to bring on some system integrated partners that we are just signing up and in the process of training to go to their clients so that we can scale more. We have some OEM opportunities, and then we have an SDK that we are also launching towards the end of the year, beginning of next year. That allows other developers to build on top of this deterministic platform and enable them to utilize language when you are interacting with computers deterministic. That's how we're going to market, and we're focusing really first in that direct sales go-to-market process and the system integrated go-to-market process is Salesforce and Salesforce clients and enabling them to get the information and then scaling that out within that account to first finance and then marketing.

ROI of Decision Velocity

SPEAKER_04 22:24

It makes a lot of sense. And you know, if you have some traction with end customers directly, then you can train the SIs on what you've done to be successful and you know what it took to get customers actually to get the benefit from the solution, right? This is where they want to they'll want to take it. I sort of want to land our episode on kind of the business benefit here. And you had frame this as a race to compress the cycle from data to decision to action. And that's sort of where we started our conversation. And I wonder, can companies place an ROI on speed of decision, quality of decision? And you know, is decision velocity a moat for firms like yours? You know, the ability to can you sell decision velocity? It's kind of an interesting proposition.

SPEAKER_00 23:15

I don't know. We're about to find out. That's the honest answer. But I think that if you think about it, you know, Jill Cates has a perspective that cycle times the information. If you can shorten that cycle time to information, shorten cycle time to decision to action and then to results. So today, the average ad hoc report in a mid-sized to large enterprise in the United States takes 2.7 weeks on average. And also in that situation, the people who can get to that report are generally executives who have access to the analyst cadre. Compare that to another organization. They can just speak or write, ask questions, it picks the data from the relevant data sources, answers back in a second. Then they can ask follow-on questions and that. That information is 100% accurate, auditable, and everyone in the organization can do that. How do you think the competitive environment will look between those two firms? One second, 2.7 weeks.

SPEAKER_04 24:27

Yeah, I think the ability to cycle through effective decision making could be a huge competitive advantage. And we're gonna have to watch Quario's success in the market.

Wrap Up and How to Connect

SPEAKER_00 24:38

We always look at the board and executive team as the people who make decisions, but in reality, most of the decisions in an enterprise are made by the line of business people and the managers that guide them. That's the majority, that's 99% of the decisions in the business. The executive team and board takes 1% of the decisions that are strategic. Everything else is the rest of the organization. So let's make everyone able to make decisions and take actions, and then we'll win. And then that company will win. And that's our goal to make our customers win.

SPEAKER_04 25:18

I love it. So, KG, if people want to connect with you or learn more about Quario, how can they do that?

SPEAKER_00 25:25

They can contact me via LinkedIn. They can go to our website, quario.com, q-u-ar-r-i-o.com. Those are the two simplest ways to contact me. Fabulous.

SPEAKER_04 25:39

Thank you. And folks want to reach me, I'm on LinkedIn slash in slash Ken Lempit. My demand generation and advertising agency for software and AI companies is AustinLawrence. We're at AustinLawrence.com. And if KG Charles Harris hasn't convinced you to subscribe to the SAS Backwards Podcast, hard to imagine what will do it. Thanks again for being here, KG, and we'll see everybody on the next episode of SAS Backwards.

SPEAKER_02 26:10

Thanks for listening to the Sass Backwards Podcast, brought to you by Austin Lawrence Group. We're a growth marketing agency that helps SaaS firms reduce churn, accelerate sales, and generate demand. Learn more about us at www.austinlawrence.com. You can email Ken Limpet at kl at austinlawrence.com about any SaaS marketing or customer retention subject. We hope you'll subscribe and thanks again for listening.