#266 Andy Kurtzig: How Pearl Uses AI + Human Experts In Professional Services

AGNTCY - Unlock agents at scale with an open Internet of Agents. Visit https://agntcy.org/ and add your support. What if AI could actually be trusted in healthcare, law, or even car repair? In this episode, Andy Kurtzig, founder and CEO of Pearl (formerly JustAnswer), reveals how his platform is solving the biggest problems in AI: hallucinations, liability, and monetization. Pearl blends large language models with real human experts across 700+ categories—so users get AI speed and human-verified accuracy. Whether you're a doctor, mechanic, or just someone with a question, Pearl offers fast, affordable, and expert-backed answers—without the risk of misinformation. If you're building in AI, working in a regulated industry, or just curious about the future of trusted automation—this episode is for you. Check out Pearl, AI Enhanced Human Expertise: https://www.pearl.com/ Stay Updated: Craig Smith on X: https://x.com/craigss Eye on A. I. on X: https://x.com/EyeOn_AI (00:00) Introduction (03:20) Why Pearl Focuses on High-Stakes Industries (06:10) How Pearl Reduces AI Hallucinations with Human Trust Scores (10:11) Pearl.com's Ideal Customers (14:14) Solving AI Liability with Platform Protection (23:27) The Three Big Problems in AI: Risk, Quality, and Monetization (29:00) Why Human-in-the-Loop Is the Future of AI (33:26) Pearl's API: Bringing Verified Experts to Any App (39:27) Building Trust in AI Through Human Validation (41:03) The Road Ahead for Pearl and Human-Centered AI

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Pearl will give you an AI answer and then you'll get an actual doctor will show up, read the conversation, read the answer and then give it a trust score from one to five. We focus on primarily text and professional services. So medical, legal, veterinary, accounting, tech support, home improvement, these kinds of areas. So and that focus yields a very different approach to how we handle things. So and that approach results in a significant improvement in quality for the user. We do, we've got 12,000 experts that are on our network. Like I said, we've got pearl.com is part of a collection of businesses.

• 0:42Speaker 00

  And across those businesses, we have about 12,000 experts that are at the beck and call of our pearl.com users. And so they're there to jump in and verify these answers. We figured out how to monetize it. And that is through real life, human expert professionals, that the human expert needs to be in the loop. People understand that a doctor costs money, a lawyer costs money, an accountant costs money, a fraction when you do it through AI and online. But people are used to paying money for that. And that creates a business model here that is both supportive, that improves the quality, reduces the risk and creates monetization opportunities.

• 1:19Craig S. Smith?

  Building multi-agent software is hard. Agent to agent and agent to tool communication is still the wild west. How do you achieve accuracy and consistency and non-deterministic agentic apps? That's where agency, A G N T C Y comes in. The agency is an open source collective building the internet of agents. And what's the internet of agents? It's a collaboration layer where AI agents can communicate, discover each other and work across frameworks. For developers, this means standardized agent discovery tools, seamless protocols

• 2:03Craig S. Smith?

  for inter-agent communication and modular components to compose and scale.

• 2:10Craig S. Smith?

  Multi-agent workflows. Build with other engineers who care about high-quality multi-agent software. Visit agntcy.org and add your support. That's A G N T C Y dot O R G. Visit them today to support high-quality multi-agent software. Go ahead, then.

• 2:34Speaker 00

  Well, it's nice to be here and happy to give my background. So my name is Andy Kurtzig. I'm the founder and CEO of Pearl, pearl.com, like a pearl of wisdom. And what we do is we take solo professionals and turn them into AI superheroes using essentially agents to do the sales and the marketing and the customer service and the account management and payments and all that for them. So they can focus on what they love, the actual human judgment expertise part and not the other stuff. And I've been doing this for sort of blending AI and humans together for roughly 20 years now. And really proud of how far we've come.

• 3:20Craig S. Smith?

  Yeah. When did you found Pearl?

• 3:23Speaker 00

  So Pearl is a subsidiary essentially of Just Answer, which I founded 20 years ago.

• 3:30Craig S. Smith?

  Oh, okay.

• 3:31Speaker 00

  So, yeah, it's a collection of brands. Just Answer, Pearl. I've got a couple other businesses as well.

• 3:38Craig S. Smith?

  Yeah. So Pearl is, what I understand, is an AI platform that combines L L Ms and a network of human experts, excuse me, to verify responses. Can you tell me about, I mean, it must be tough in this environment to compete with Chat G P T or Perplexity. Can you tell me how you compare to those guys and how you manage that?

• 4:10Speaker 00

  Of course. So first of all, we're very focused on the professional services sector. So we're vertical, they're horizontals, right? They're doing everything from coding to video and images and and everything in between. We focus on primarily text and professional services. So medical, legal, veterinary, accounting, tech support, home improvement, these kinds of, kinds of areas.

• 4:35Speaker 00

  So and that focus yields a very different approach to how we handle things. So and that approach results in a significant improvement in quality for the user, for the consumer. So specifically, we're 22% more accurate than ChatGPT and Google Gemini in professional services.

• 4:59Craig S. Smith?

  And professional services, it runs the gamut from

• 5:05Craig S. Smith?

  legal to healthcare to, I mean, how how broad is vertical?

• 5:12Speaker 00

  Yeah. So if you take all the subcategories, so even within medical, you've got everything from oncology to dermatology and endocrinology. If you take all the subcategories, you put them all together, it's roughly 700 different different categories that that we that we focus on and and they all have some very similar characteristics. Number one is that quality really matters, right? And and humans are a key element of of delivering that quality.

• 5:43Speaker 00

  Obviously, in in medical or legal, for example, making a mistake is extremely

• 5:49Craig S. Smith?

  Yeah.

• 5:49Speaker 00

  unfortunate and costly and and and and, you know, not right, not a good place for lots of hallucinations.

• 5:57Craig S. Smith?

  Sure. And how does that work? Then you have this network of human experts. Someone asks an LLM a question. Oh, just describe how it works.

• 6:10Speaker 00

  Yeah. So so it's fairly straightforward. You go to pearl.com, you type in your question. It works kind of like a like a ChatGPT or Gemini or Perplexity. At the beginning, you you you type your question and then.

• 6:27Speaker 00

  It'll go back and forth with you typically, more like, a little bit more like some of the deep research tools out there do. And qualify and actually take, get rid of the need for fancy prompting, because we find that's a big source of hallucinations is, is the prompt that the average person puts in isn't great, and so then the, the output isn't great. So we take care of all that. So you just put in, in your language, what you're trying to do, I have a rash on my foot, or whatever it is. And we, we will go back and forth if you clarify the question, just like a doctor in a live environment would. And then we actually give you the, the, the AI answer from pearl.com. That's just the beginning. That's, that's, that's what's 22% more accurate.

• 7:10Speaker 00

  And more importantly, actually, than the 22% more accurate, it's 41% less wrong. So when you're dealing with medical, legal, these kinds of cat, accounting, these kinds of categories, being wrong is what you're really trying to avoid, because that's where the catastrophes happen. Um, and, and so that's the, that's the first step. Higher quality AI LM answer. But then the, the, the additional twist here is, we will then have an actual human expert verify the accuracy of that AI response, of Pearl's own response. So in that case, with your rash on your foot, Pearl will give you an AI answer.

• 7:53Speaker 00

  And then you'll get, an actual doctor will show up, read the conversation, read the answer, and then give it a trust score from one to five. So that you then know, okay, can I trust this thing or not? And if you can, great, have a nice day. And if you can't, well, then you gotta keep digging.

• 8:12Craig S. Smith?

  Yeah, you know, I've, there, there, a lot of these systems now that have an, an LLM verifier that, that looks at the answer from another LLM and scores it. Is any of this automated? And if not, how do you do it in real time? I mean, do you have? Experts in each particular field on call. You know, twenty four seven or yeah, how do you, how do you manage that?

• 8:47Speaker 00

  We do. We've got twelve thousand experts that are on our network. Like I said, we've got. Pearl dot com is is part of a a collection of of businesses and and across those businesses, we have about twelve thousand experts that are. At the beck and call of our pearl dot com users. And so they're there to to jump in and and verify these answers and read them and. And give the trust course. And then if you want to as a as a user. You can then choose to pay and then actually end up on a phone call or a live chat with the doctor about him for as long as you want. So that's that's kind of step three of the. The the experience and that's where that's where the money comes as well. That's where the monetization happens. So the AI answer is free.

• 9:32Speaker 00

  The the the trust score is free from the human expert, but then if you want to talk with them, then you can pay a monthly fee and talk to them all you want.

• 9:42Craig S. Smith?

  And then you pay a monthly fee. It's a not per call.

• 9:45Speaker 00

  Yeah, that's right.

• 9:47Craig S. Smith?

  Okay, and uh, so who are the users in this? Are they casual users that want one answer?

• 9:57Craig S. Smith?

  For one thing, or is it's other medical professionals or other lawyers, for example? Or paralegals that, I mean, is there, is it more a B to C or a B to B product?

• 10:11Speaker 00

  It's interesting. We launched it as a B to C product. And we're finding a lot of power usage out of the B to B group. So like you said, lawyers or or doctors or. Or mechanics or accountants that just need help. So I'll give you a funny story. In the mechanic category, we're seeing a lot of mechanics. That might be in a small town like the middle of Alaska or something and and they're the only mechanic in town. And so you can imagine they might know how to deal with a Ford or a or a or a Chevy, but then somebody brings them a Volvo or a BMW or a Saab or a. Whatever it is, they don't know how to repair those cars, but they're the only mechanic in town.

• 10:55Speaker 00

  And so they turn around and start using us to to be able to to service those cars as well.

• 11:03Craig S. Smith?

  Well, that's fascinating. So, and how quick is the response? I mean, if I'm a mechanic among your experts.

• 11:13Speaker 00

  Yeah.

• 11:13Craig S. Smith?

  Um, I, is it's, um, through my smartphone that I get a, an alert?

• 11:19Speaker 00

  Yeah.

• 11:19Craig S. Smith?

  I look at it and, um, and how much time does it generally take to score the, the LLM response?

• 11:28Speaker 00

  It usually takes a few minutes. It's not very long. Yeah.

• 11:33Craig S. Smith?

  Wow. Yeah. Uh, and, and then you, uh, pay the expert for their time and, and, and take a, a commission and in fact, is that right?

• 11:47Speaker 00

  Yeah. Well, so for the, for the verifications, we pay the expert for their time and, uh, and then give it to the users for free. So, so we're eating that costs in the hopes that some customers will choose to, to then become subscribers.

• 12:01Craig S. Smith?

  Yeah. And what is the subscription fee of?

• 12:06Speaker 00

  It's about, it's about 30 bucks a month.

• 12:09Craig S. Smith?

  I see. Yeah. So similar to, to any of the, Jeff, the LLM.

• 12:15Speaker 00

  Yeah. Yeah. Except for that includes actually talking on the phone or on chat live with a human doctor or lawyer.

• 12:21Craig S. Smith?

  Right. Yeah. Well, that's fascinating.

• 12:23Speaker 00

  Yeah.

• 12:24Craig S. Smith?

  Yeah. Uh, and, and on this back and forth you talk about, uh, so you ask it a question. Uh, Pearl asks these nuanced follow-up questions to better understand your intent. Um,

• 12:42Craig S. Smith?

  and, and are the, have you refined? I mean, fine tuned the models that you're using with, uh, answers collected from these, uh, experts?

• 12:57Speaker 00

  Of course. So that's a key part of what makes it 22% more accurate than, than ChatGPT and Google Gemini is, is, is the rag of, of all of our data. So we've got about 30 million past conversations between doctors and, and, and, and, and their customers and lawyers and the customers, et cetera, in our database that we can then use to increase the quality of the, the answers.

• 13:24Craig S. Smith?

  Yeah, and that fine tuning, what's the base model? Or do you use multiple L L Ms?

• 13:33Speaker 00

  We use multiple L L Ms underneath as well. So we've got, we've built our own eval system as well, because we've been doing this for so long that that that that evaluates the various different models for various different purposes. And so every time you use Pearl, you're using at least two different L L Ms are happening behind the scenes. It's all invisible to you as the user, but but at least two different L L Ms are being used behind the scenes.

• 14:02Craig S. Smith?

  Yeah. And so the there's, you know, if you're dealing with healthcare or legal, there's a potential liability issue. How do you how do you handle that?

• 14:14Speaker 00

  That is an excellent question. So so let's start with with how do the how do the L L Ms handle that? How does Chat G P T handle that? How does Google Gemini, Perplexity, et cetera, handle that? And the short answer is they don't, and they can't, and they're liable. And we've got a bunch of data on that, everything from some pre court justices to to law journals to even consumers and their intention. We did a survey recently and asked consumers if if an L L M was wrong. You know, what would you do about it? And forty percent said they would consider legal action against A I companies. And and fifty percent of Americans believe the A I platforms are legally responsible for the answers they provide.

• 14:57Craig S. Smith?

  Yeah.

• 14:58Speaker 00

  So so that's a giant flaw in the foundational models, big game plans. There's actually three big flaws. That's one of the three big flaws. We can chat about the other two. If you like, but we'll focus on risk for now.

• 15:14Speaker 00

  Huge flaw. I mean, they're starting to see the pain of that, right? You've seen the the lawsuits, that kid that killed himself because the A I told him to. And and all these things. I mean, it's just tragic. And the family of that kid and the families of some others that have been like given hallucinations, quote unquote, are now suing Google and and some of these other companies. And I think they're gonna win. They don't have an argument against it. Now, with all that said, there is this thing called Section 230. It's an act of Congress that has protected Google and Facebook and many of these other companies from that exact kind of lawsuit in the past. But there's a key, a key way that Section 230 works.

• 15:59Speaker 00

  It works for platforms and a platform or a marketplace is where somebody might come, like even on Google, you go to Google, you get a billion search results, you click on with some of one of them and you end up on some other third-party site. And then that third-party site is the one that's then giving you the answer and that's how they deal with it. Google can say, "Section 230, we're just the platform, you can't sue us." And that's worked like a charm for them for years.

• 16:26Speaker 00

  Until now, where they're not doing that, they're not sending you to some other site, they're not even attributing any other sites most of the time. They're just telling you, "You're going to be fine, don't worry about the cancer," or whatever it is that you're asking it about. And they're liable for that.

• 16:42Craig S. Smith?

  And when you say they're not going to a third-party, it's because they're letting their LLM answer.

• 16:47Speaker 00

  That's right, that's right. And they often don't even know the source. It just, all that data just kind of got sucked in and then they give you the answer and they don't even know if it's true or not. And they also don't know where they got it. And so they don't even attribute it. Not that that would solve it, but, but, you know, they're not even close to solving it. And so that's the, the piece where Pearl actually steps in and can, can really help out. So the interesting thing about what pearl.com is doing from a liability perspective is, yes, we have the AI answers, but then we're, we're very forward with the human expert verifications and conversations, which turns us back into a platform again, which gets Section 230 protection again, and, and it solves that big problem.

• 17:34Craig S. Smith?

  Building multi-agent software is hard. Agent to agent and agent to tool communication is still the Wild West. How do you achieve accuracy and consistency and non-deterministic agentic apps? That's where AGNTCY comes in. The AGNTCY is an open source collective building the Internet of Agents. And what's the Internet of Agents? It's a collaboration layer where AI agents can communicate, discover each other and work across frameworks. For developers, this means standardized agent discovery tools,

• 18:17Craig S. Smith?

  seamless protocols for inter-agent communication and modular components to compose and scale multi-agent workflows. Build with other engineers who care about high-quality multi-agent software. Visit AGNTCY.org and add your support. That's A-G-N-T-C-Y dot O-R-G. Visit them today to support high-quality multi-agent software. I see. And the liability, if there is a wrong answer, does that extend to the experts or are they protected by 230?

• 19:02Speaker 00

  No, so the experts are the ones giving the answer. And so they're the ones that are held liable in this case. But then they've got their own insurance and they've got their own ways of dealing with that. They know how to deal with medical malpractice or whatever it is. That's something that they're used to and have protections for.

• 19:22Craig S. Smith?

  Yeah. And have you tested how uniform are the answers across experts? That's something I've always wondered.

• 19:35Speaker 00

  Yeah. You know, it's interesting. You kind of...

• 19:40Speaker 00

  In some ways you want uniformity, but in other ways you don't, right? So when you've got a, a, a medical condition, you don't want everybody telling you, sorry, you're out of luck. Have a nice life. You want a bunch of different angles and a bunch of different perspectives and points of view. And sometimes the person says, sorry, you're out of luck. Have a nice life. The next one says, oh, this is new thing over here. Have you heard of that? You might want to try this. I don't know if it'll work. And so some of the, the differences are actually useful, especially when you're dealing with, with tricky diagnoses and things like that.

• 20:08Craig S. Smith?

  Well, that's interesting. So if I have a medical question, I go on Pearl, I ask it, I get an LLM answer. Do I have to prompt it for the trust score?

• 20:22Speaker 00

  Nope, it'll prompt you. It'll say, hey, you know, it'll prompt you and sometimes it'll even just automatically do it for you, depending on the situation.

• 20:30Craig S. Smith?

  Right. And then I get the trust score and do I have to ask for the, to talk to the expert then?

• 20:38Speaker 00

  Yeah, if you actually want to have a phone call or a live chat with the expert, you have to ask for that and you have to sign up for the subscription.

• 20:45Craig S. Smith?

  Sure. Yeah. But can I then talk to two experts to get a second opinion if I don't like the first one?

• 20:53Speaker 00

  Yeah, yeah. With the subscription, you can do, absolutely do that. And that's one of the beauties of this is, is you can get multiple opinions really easily, right? The part of the reason a lot of people don't get second or third or fourth opinions is because it's so hard. You got to make an appointment for three weeks from Tuesday. You got to drive down there and you show up and you know, and they're late and whatever else happens. You know, now you can sort of spam, bam, bam, get three different opinions very quickly.

• 21:17Craig S. Smith?

  Yeah, actually. That sounds amazing. I mean, did you consider a charging per session or per question? Because a lot of times people.

• 21:34Craig S. Smith?

  Only want one answer. They they don't necessarily want a subscription or is you think the subscription, can you cancel, you know, sign up for a month and cancel immediately and then you've you've spent thirty dollars to get an expert opinion, which is extremely uh, cheap. Um, I mean, how does that work?

• 21:58Speaker 00

  Yes. So first of all, it's very easy to cancel. So yes, you can just sign up, get your one or your two or your three opinions or whatever you need. You got a whole month to do it then, and then you can cancel and only pay thirty bucks for all that. And yeah, but what we find is people really like the ability to sort of ask a second and a third opinion. And a lot of these things kind of end up being maybe chronic issues. And so you want to be able to ask over time as well. And this is a way that we can price it lower. So obviously, if you just, you know, without insurance, showed up at a doctor's office and said, hey, I want a one opinion, they're going to charge you hundreds of dollars for that, plus the drive time and everything else you have to deal with. Now we're charging thirty and you can get multiple opinions that way.

• 22:44Craig S. Smith?

  Yeah, yeah. Um, okay, the um,

• 22:50Craig S. Smith?

  it it the the um, I'm sorry, I'll cut this out. I just had a message pop up that knocked my train of thought out of skew.

• 23:01Speaker 00

  No problem.

• 23:05Craig S. Smith?

  Uh, just I'm reading some of these notes.

• 23:15Craig S. Smith?

  Yeah, I'm seeing this note about why Pearl is the first truly monetizable AI search prob- prob- platform. Can you, can you go through that?

• 23:27Speaker 00

  Yeah. So let me, let me frame this in a context and then I'll answer that question specifically. So

• 23:36Speaker 00

  I believe there are three fundamental problems with the foundational models out there, the chat GPTs and then Google Gemini's and perplexities and such. And we've already talked about one, which is risk. They're they're they're liable for huge amounts of risks. I mean, they're going to end up with billions and billions of dollars worth of lawsuits from everybody that that they hallucinated to and caused damage to. And and it's not a big deal when you're writing a poem, maybe, but for the family dinner party, but when you're dealing with medical, legal, these kinds of categories, it is a big deal and they're not taking that seriously enough. And so they're giving out wrong answers and they're hurting people and they're hurting pets and they're hurting people's legal situations, et cetera.

• 24:20Speaker 00

  So so risk is a big one.

• 24:25Speaker 00

  Second issue of three is quality, the underlying quality, right? So we're seeing hallucinations like crazy. And in fact, with these new versions of models that just came out, they're actually hallucinations went up, not down. This is this is a fundamental part of how these these foundational models work is that they hallucinate. That is just embedded in how they they structurally work. And so those hallucinations, again, are a big problem and nobody's got a an idea on how everybody's got lots of ideas, but nobody has a clear path on how to how to get past the hallucinations. Things just seem like they're just a part of these systems. And then the third, of course, is monetization.

• 25:07Speaker 00

  We're seeing in the news, left and right, these not only these these AI companies, but but many companies in and around the space are losing money hand over fist, right? They're spending this much on GPUs from Nvidia and there's they're getting this much in revenues from, you know, if they're getting any revenue at all, you know, Nvidia is making a lot of money off of this whole thing, but but most of the foundational, all the foundational models and and and big players are losing lots of money on on on AI today. And so those are the three big fundamental problems with the with the foundational models. It's quality, risk and monetization. You asked specifically about monetization, so let's talk about that.

• 25:49Speaker 00

  They don't have a good monetization method yet. Charging 20 bucks a month kind of works for early adopters like you and me and and small businesses and stuff. But but that's not the model that's going to end up winning, you know, something more like what Google does with a free model is what's going to end up winning to reach the masses of consumers. And we're seeing the free models out there. That's lovely. But it's they're not figuring out how to monetize it yet. So where are the ads going to go? And what are you going to do when when when you ask what's the best tennis shoes to buy? And the answer is Nike, but the but the advertiser is Adidas, right? What are you going to do? Right? And and so they don't know what they're going to do with that.

• 26:35Speaker 00

  And and it's not like kind of Google where you can just say, well, we kind of suck anyway. Here's a billion links and a billion ads and you just sort of you figure it out. They're trying to tell you an answer. And when you start getting into the answer business, it's tough to to to both give a great quality answer and

• 26:55Speaker 00

  put ads in there.

• 26:56Craig S. Smith?

  Yeah.

• 26:57Speaker 00

  You can imagine asking for the score of the, you know, NBA game. And, you know, one model decides to to say, well, before I tell you the score of the NBA game, let me tell you about the all new Toyota Tundra. Right? Nobody would go to that model. They'd go to the next model down the list. And so there's there's a lot of question marks still about how to monetize these things. And we figured out how to monetize it. And that is through real life, human expert professionals, that the human expert needs to be in the loop. People understand that a doctor costs money, a lawyer costs money, an accountant costs money. It's a fraction when you do it through through through AI and and and online.

• 27:39Speaker 00

  But people are used to paying money for that. And that creates a business model here that is both supportive, that improves the quality, reduces the risk and creates monetization opportunities.

• 27:50Craig S. Smith?

  Yeah. And before I go on to what's going to happen to these LLM based search models or engines or services, whatever. How do you so I I asked Pearl about.

• 28:11Craig S. Smith?

  A bump on my nose that may be malignant, you know. It gives me an answer. I get a trust score, but I want to talk to an expert. Uh, this is all within my subscription, I imagine, but once I have that,

• 28:34Craig S. Smith?

  excuse me, that expert on the phone, uh, you know, he answers the question. I say, yeah, but you know, I have ringing in my ears. What should I do about that? I mean, you can imagine that the conversation with the expert goes on for an hour is suddenly this person has an expert on the line and can ask all the questions that they want to ask. How do you prevent that?

• 29:00Speaker 00

  We don't prevent that. We love that. We're trying to help people. That's at the end of the day, what our mission is, is to help people and and we're happy to help people. Not only do we, do we help people like that and that kind of situation with that doctor, but we have all kinds of specialists. I mentioned, we have 12,000 different experts. And so sometimes the person that's going to talk to you about the bump on your nose is a different person than the one that's going to talk to you about the ringing in your ears. And we're happy to to connect you with the, the ear doctor to talk about the, the latter and, and, and the dermatologist or whatever it is to talk about the former or the ENT.

• 29:33Craig S. Smith?

  And that's all within the monthly subscription?

• 29:36Speaker 00

  Yep, yep. Yeah, it's amazing.

• 29:38Craig S. Smith?

  That is remarkable. Yeah. Uh, and, and so what's going to happen to AI search?

• 29:46Speaker 00

  Well, I think they're going to have to evolve. So I mean, the, the quality, I think we all have to get used to the fact that the this is about as good as the quality is going to get. It's, there's going to be lots of optimizations, but the hallucinations are, are endemic to the, to the way these things work. And we're not going to get to 99% accuracy that like you would need for, you know, you need even more than that for healthcare and legal and things like that. You have to assume, I mean, our estimates and professional services are that they hallucinate about 37% of the time. So you have to sort of recognize going to hallucinate. And that means there's certain jobs that it can do and certain jobs that it can't do. And we just need to sort of get good at

• 30:29Speaker 00

  recognizing what kinds of jobs to use it for, what kind of jobs not to use it for. So that's on the quality side. I will say, we should expect that the prices will continue to come down. So that the competition and, and, and you know, these, these L M's are becoming commodities, you know, between all the open source models and deep seek and everybody else coming into the market. Uh, the, the, the cost to the users will continue to, to fall.

• 30:56Craig S. Smith?

  Yeah. Uh, what are, are there cases where you think A I is, can be trusted more than, than others?

• 31:06Speaker 00

  Yeah, where you're an expert, right? So I'm an expert about the email I want to write to my friend. I'm an expert about, uh, you know, what kind of logo I want for my business. I'm an expert on, you know, these kinds of things, anything that I can evaluate properly. I, you know, he said, so if it gets it wrong, I don't like that letter. I don't like that email. I'm going to change this, this, this, this, and this. I don't like that logo. Try again, try again, try again. Right? If I'm a good judge, if I'm an expert myself in that topic, and I, and we're all experts at many things, right? Then it's fine, because we can judge it, we can know how, where to use it, where not to use it. The problem is when we're not an expert,

• 31:49Speaker 00

  when we use this for medical questions or legal questions or for veterinary questions, and A I tells us something. And by the way, it tells us with a hundred percent confidence every time, oh, no, don't worry about that cancer. You're going to be fine. Just eat some rocks and, and it'll go away. You know, oh, okay. You know, I mean, that's, that's what these things are doing out there. But, but in the absence of that knowledge, right? I'm not an oncologist. I wouldn't know. And so I get this information and I trust it. And that's where you got to be careful. That's where the big mistakes happen. And that's where you need a human in the loop, a real life human expert.

• 32:25Craig S. Smith?

  Are there cases where the human expert says, "Look, this issue is too complicated or really requires a face-to-face meeting with a professional?"

• 32:42Speaker 00

  Of course, of course. Yeah, I mean, you can imagine all kinds of scenarios, right? A surgery is required on your pet. I can talk to you about what that surgery might look like, but I'm not doing a surgery online, right? I can't go represent you in court, you know, on Tuesday, for example. I mean, these kinds of things, you need to go face-to-face.

• 33:01Craig S. Smith?

  Yeah. So do you think this is

• 33:07Craig S. Smith?

  the model that others will eventually adopt? And what about

• 33:15Craig S. Smith?

  corporate executives that are pouring money into developing their own enterprise L L Ms?

• 33:26Speaker 00

  Let's see. So in terms of the, is this a model that others will adopt? Yes, we're big believers in that. In fact, we've built an API. So we've got a product called, we call it internally Experts as a Service. And you can go to pearl.com slash API. And you can stream our experts right into your application through our API and do that. And just like you would stream it, OpenAI into your application, you can stream real-life human experts, doctors, lawyers, et cetera, right into your, your application or for your use case. And so we're big supporters of that. And we do think this is a future. The combination of the best of what humanity brings, combined with the best of what AI brings, gives you the best possible solutions. So that's, that's where I think the world's going.

• 34:11Speaker 00

  That also that solves the quality problem, that solves the risk problem, and that solves the monetization problem. It's the hat trick of the foundational, fundamental problems with foundational models. In terms of enterprise, I think everybody's doing it a different way. I think, you know, some are doing it better than others. Most are doing it poorly today.

• 34:36Speaker 00

  But but you gotta be doing it. If you're not at least playing with it, you're you're gonna get left behind. So I think that's kind of what's causing people to be playing with it and trying to do things with it. But I think we're learning a lot very quickly about how to how to make these things useful.

• 34:49Craig S. Smith?

  Yeah, the API is fascinating. So if if I built a

• 34:57Craig S. Smith?

  chatbot

• 35:01Craig S. Smith?

  to talk about the history wherever, you know, GPS based chatbot that would tell you the history of wherever you are. Do you have the experts on call to to back up answers? It's it's something I'm personally interested. I love history.

• 35:22Speaker 00

  Yeah.

• 35:24Speaker 00

  Yeah. So so

• 35:28Speaker 00

  so the experts, we have to look through the expert category to see what kinds of experts you want. So so we've got 700 different categories. But but the big buckets are medical, legal, veterinary, accounting, tech support, home improvement,

• 35:44Speaker 00

  appraisals, right? We got a lot of categories, so we'd have to go through the categories and see, or any of these match up what you're you need.

• 35:51Craig S. Smith?

  Yeah. Well, actually, that raises another question. How does

• 35:57Craig S. Smith?

  the LLM know which expert to route the question to?

• 36:03Speaker 00

  There's multiple different ways. In the API, are you talking about or in in Pearl.com?

• 36:09Craig S. Smith?

  Well, in Pearl.com to begin with, yeah.

• 36:11Speaker 00

  Yeah, in Pearl.com, we use AI, of course, to to route it to the right one.

• 36:16Craig S. Smith?

  And the API, then?

• 36:18Speaker 00

  The API, we can do it either way. We can either use the use AI to route it to the right one, or you just tell us, right? I'm a medical site and I only want doctors. This is going to go to doctors and you can pick the specialty or not when you when you send in the API call.

• 36:33Craig S. Smith?

  Yeah. And so this API, I mean, that's really interesting.

• 36:38Craig S. Smith?

  A hospital on its portal could have

• 36:44Craig S. Smith?

  A white labeled question chatbot.

• 36:48Speaker 00

  Exactly.

• 36:49Craig S. Smith?

  That that routes through the API to your experts.

• 36:53Speaker 00

  Exactly. And there's multiple use cases for a hospital. There's there's that sort of B to C kind of consumer application. Hey, you know, if you're really worried about this and you want to talk to a doctor about it at 3 a.m., we got one ready for you right now to do that. So that's one application. They can do that either with the AI like we do with Pearl. Combined with the human or just straight to human if they want to do it that way to the human doctor. The other use cases we're finding in in healthcare environment is just quality is so critical in healthcare. If you're a hospital and you're even playing with your own LLM. Quality is almost always the first question that that they're asking and and trust and how do we ensure

• 37:39Speaker 00

  that that that we're we're using LLMs, but we're doing so in a trustworthy quality fashion. And that's where they can stream in live doctors to to verify things even for internal uses. Right? So even if you're a nurse practitioner and you want to use this, or even if you're a doctor and you want to use this. Right? How can we up the qualities by the time it even gets to that doctor so that that it's more trustworthy than just hallucinating LM?

• 38:08Craig S. Smith?

  Yeah. And you were saying that your platform.

• 38:15Craig S. Smith?

  If I'm a hospital and I want this functionality, but I don't want to leave it to your experts. Can I use the API and and then have it route to my experts?

• 38:33Speaker 00

  We haven't done that before. I mean, usually people would you would have their own system for that. Right? So so if your plan is routed to your own experts, you may or may not need that platform. You can certainly use Pearl for the AI part.

• 38:48Craig S. Smith?

  Yeah.

• 38:48Speaker 00

  And then and then try to do your own thing with your own experts at that point. So that's easy to do. But but.

• 38:57Craig S. Smith?

  Yeah.

• 38:57Speaker 00

  Yeah, I guess I guess that's doable in that way.

• 39:00Craig S. Smith?

  Yeah, I'm just thinking that there may be organizations that that don't have the want to spend the money. You've already created the system.

• 39:10Speaker 00

  Right.

• 39:11Craig S. Smith?

  Yeah. And then

• 39:15Craig S. Smith?

  can you talk about you? You've done some research into AI accountability and trust. Can you talk about some of the findings?

• 39:26Speaker 00

  Of course. Yeah. So we asked, you know,

• 39:31Speaker 00

  whether consumers would pay for AI if it meant better, more accurate answers. And forty two percent said, yes, they would. Another another thing that we learned was that nearly forty seven percent of Americans would trust AI more if humans validated its answers. Which is obviously part of what led us to this path. And then we already talked about some of the data about the legal liability part of of things and how at least Americans would,

• 40:04Speaker 00

  do hold these L M liable and and would forty percent say they would consider legal action from AI companies if if they were given wrong answers.

• 40:13Craig S. Smith?

  Yeah. And how are you? Is this? I mean, it seems the the the market would be endless depending on your distribution. And you have just answers, which has a large user base, but

• 40:31Speaker 00

  Right.

• 40:32Craig S. Smith?

  What are you doing to to

• 40:36Craig S. Smith?

  get this out there? I mean, are you? Are you primarily looking at

• 40:43Craig S. Smith?

  integrating with other

• 40:46Craig S. Smith?

  platforms and services, as I said, with hospitals? Or maybe law firms? Or are you primarily just marketing the direct service through pearl.com?

• 41:02Speaker 00

  That's where you come in, Craig.

• 41:06Speaker 00

  So we're doing, obviously, we have the API so that people can use it and whatever creative ways they can imagine. And then we are also a marketing pro.com directly out in the market and doing that so that we can be on the cutting edge and learn about how consumers are using it and what they want and don't want and try to make this. Eat our own dog food so that when people are using our API, for example, they're getting the best of the best and the latest latest thinking on this stuff.

• 41:34Craig S. Smith?

  And how many users does Just Answers have?

• 41:39Speaker 00

  In terms of traffic, we were just named SimilarWeb's fastest growing site in the world last year, number one. ChatGPT number nine, by the way, on that.

• 41:50Speaker 00

  We've got about 60 million visitors a month.

• 41:52Craig S. Smith?

  Wow, yeah. Yeah. Yeah, this is fascinating. I mean, I'm definitely going to use it.

• 42:02Craig S. Smith?

  You know, I don't know if I'd want a subscription running for a year, but you know, it's it's inexpensive enough that I could dip in and out. Now, when I have questions and I love the idea that you could connect it to your own site. By API and how do you charge there? Is it a usage or is again just a monthly subscription?

• 42:26Speaker 00

  That that one's typically a usage. Yeah, they've got their their their pricing. It depends on the size and scale and those are kind of custom prices at the moment, but. But yeah, it's not a per usage per question basis.

• 42:39Craig S. Smith?

  Right. If if someone's using the API and can't find. A category to fit their their vertical. Can they talk to you about developing an expert cohort for that vertical or how how do you as you grow? I imagine you're adding. Sort of broadening the categories.

• 43:03Speaker 00

  Yeah, but we do have 700 categories, so we got a lot to start with and it and it is a lot of work to get one set up. So it's not to sort of, oh, you know, wouldn't it be great if you had this and then the next day we've got that you got to build a liquid network. I mean, in order to to be able to have the response times that that people demand, you know, a few minutes kind of thing. You got to have a lot of a lot of professionals on the other end of that.

• 43:28Speaker 00

  And in order to have enough professionals, you got to have a lot of consumers too to sort of build up the network effects and build up the demand. So, so it is a big endeavor to add a new category and, and we do so thoughtfully and carefully. But, but we, but we've done that over the last 20 years. We've built up about 700 categories.

• 43:48Craig S. Smith?

  Yeah, um, the, um. You know, I'm just thinking, I, I spent my career at the New York Times and the Times owned about.com for a long time. Who bought about.com? Do you remember? It wasn't you guys, huh?

• 44:06Speaker 00

  It wasn't us. Yeah, that was so, I remember the deal very well, but, uh, who bought them?

• 44:14Craig S. Smith?

  Yeah, well, that's, that's okay. I was just, just curious whether it was you guys, but, um, it, in the integrations, these API integrations, do people generally uh, pass it along to, uh, users as a loss leader just as a service? Or are they, um, adding a margin in charging? So it covers their cost to you?

• 44:42Speaker 00

  I've seen both. So the, and I guess the obvious example is where they, when they do the latter, where they, they, they, they charge a premium and then they, they, they keep the difference and, and and, and that's, so that's their business model. And that's a perfectly reasonable way to do it. The other thing we've seen that's creative and clever is, is what we've seen with, with one of the giant, giant secondary markets for, for car parts. You can guess the name of the company. Um, uh, uh, we have a deal with them where they make it free for their consumers to help them select the correct car parts for their vehicle.

• 45:23Speaker 00

  So the problem that they had was a lot of people are coming to the, to this, this giant site and thinking they knew the part that they needed for their, to fix their car. And then of course, many of the times it was the wrong part. And so they'd return them.

• 45:39Speaker 00

  That was a big cost. And the other problem was, a lot of people would just be uncertain. They would get into the, to the conversion funnel. They'd, they'd want to buy this part, but they're, I'm just not sure enough that this is the right one. Screw it. I'm going to go down to Pep Boys or whatever the local shop is in Autozone and, and buy it there. And so we, they, they brought us in to have the mechanics do that task for their consumers. So here's the problem. Which part do I need? The mechanic, the Ford mechanic can tell you, oh, it's this part and, and for your particular car. Now you can buy with confidence. So their conversion rates went up. So they made more money by selling more parts and their returns went down.

• 46:22Speaker 00

  So they saved money on the other side too. So that ended up being a very good ROI.

• 46:28Speaker 00

  As well.

• 46:29Craig S. Smith?

  Yeah, I'm, I'm really fascinated by this.

• 46:34Craig S. Smith?

  Is the pricing, the usage pricing on the API, would it be

• 46:41Craig S. Smith?

  possible to build

• 46:44Craig S. Smith?

  a site that answers questions using the API that charges less than the $30?

• 46:53Speaker 00

  It would be possible. It would be. Yeah.

• 46:56Craig S. Smith?

  Yeah. Is that something you guys have a problem with or encourage or?

• 47:02Speaker 00

  It depends. I mean, what, what, it depends on the partner. I mean, I think there's, you know, the reason that that might work is if, if a partner had a big brand, for example, you know, then, then they might be able to, to pull that off because they're paying less than customer acquisition costs and things like that. So that might work for them.

• 47:21Craig S. Smith?

  Yeah.

• 47:23Craig S. Smith?

  Okay. Is there anything I haven't covered that, that, that you want to say?

• 47:29Speaker 00

  I don't think so.

• 47:33Speaker 00

  Just, you know, I appreciate the work that you're doing and I appreciate our time today.

• 47:38Craig S. Smith?

  Yeah, yeah, no, I'm, I'm thinking of ways I could integrate this into my, into my site or into my podcast because it's Yeah, it's an amazing idea that I really haven't heard anywhere before. So I would think that this will be quite successful. Like many things, it depends so much on on exposure and distribution.

• 48:07Speaker 00

  Just appreciate your time, Craig. This has been fun.

• 48:09Craig S. Smith?

  Yeah, okay, great.