Introducing "Claire" - Winner of the Dreamforce Dreampitch Competition | Salesforce

Yes.

All right, ladies and gentlemen, let's welcome.

Can I ask you a question? Can we kick it up a notch and get a little rough with them?

Or is that Chris Sacca? You can do whatever you want, kiddo. Let it go. Let it go. Be nice, but let it go. All right, Misha, Marta, Claire, let's go. Ladies and gentlemen, welcome them up.

So, hey, everyone. I'm Marta.

And I'm Misha.

And we're the co founders of Claire Clair is a chatbot for testing consumer products. So we help brands and retailers identify their best sellers and their worst performers before they commit to any production decisions.

Yeah. So we're already fortunate enough to be working with some of the largest retailers in the country. Some of our first customers are Kohl's and Target, and we're working with fashion leaders like Rebecca Minkoff, Ralph Lauren, and Rent the Runway.

And we're solving a huge problem for them. That problem is that more than 50% of new products fail.

So.

So of the $24 trillion retail industry, $1.1 trillion are being left on the table every single year due to poor product decisions.

And that's not really the case in software, right? Software doesn't have this problem because software ships instantly. And software companies have developed whole methodologies like Lean, Agile, Scrum to ship and iterate product fast. But for testing physical products, Enterprises rely on 100-year-old technology is pretty much just surveys and focus groups. And these methods are just not accurate. So it's no wonder that most new physical products that hit market fail.

And that's exactly why we created Claire. You can think of Claire as a giant focus group over chat that's actually scalable and accurate. So here's how it works. Users come to our chat on Facebook messenger and provide their feedback on new products. We take that information, pass it through our proprietary data models, and tell our brand partners how their products will perform. Our chatbot is also entirely built on Heroku. So let's take a look at a live example. Let's say that we're working with a brand called Scully, which makes awesome western shirts. Scully sends Misha an email with a link which drops him directly into the chat. All he has to do is just press get started to to begin.

So I press get started, and Claire sends the first message. So saddle up and tell us which cowboy shirts you like best. How about it? Yeehaw. Of course I'll press that.

Am I allowed to invest already? Sorry.

All right, partner, would you buy this shirt? And so then I can take a Closer look at the model and just see if this shirt resonates with me. And I think in this case I'm going to say, yes, sir. Yeah, I'm going to take it. So I can say whether it's a style, color, or price, or I can have a conversation with Claire, which a lot of people do. In this case, it just looks good on the model. So I'll just tell her that looks good on model Perfect.

It does. Really does.

Cool.

Thanks for letting us know. Congrats. You won your first point. Do a little dance. So we keep customers engaged and reward them for their feedback by rewarding with points and relevant gifts.

So you can see why our completion rates are literally 10 times higher than surveys. We're also five times more accurate than traditional market research at predicting product performance because we pair our chats with data models on the back end.

And the market we're going after is quite big when you think about the sheer number of physical products released each year. So there are 400 million products that go into market and over half of them fail. We charge from 100 to $250 per product. So you're looking at a total addressable market of $20 billion. And we'll win because we are product development and data science experts. Before this, Marta worked at Mondelez, a giant snack company, where she created the department that led new product introductions for brands like Oreo, trident, Cadbury across 165 countries.

And Misha is the data expert. He was the top physics PhD student at the University of Chicago and consulted for companies like Microsoft on the future of enterprise data. So if you want to join our mission of making physical products as easy to test as software, come say hi to Claire. Thank you.

Thanks, guys.

Right on.

That was a great pitch, by the way. Thank you for that. I wanted to actually talk about. You touched on the 50% addressable market and you said this is the market opportunity. How much of that do you think addressable to you specifically in your company?

Yeah, so basically that was for all physical products. We're currently attacking apparel and home. That's kind of the place where we've done work. And that's about a quarter of that. So that's a $5 billion market.

And how many users do you think you need to actually kind of acquire to get to, let's say, profitability? How many users? Customers? And also maybe you can talk about the scale in terms of your overall attraction.

We're already profitable, Ramen profitable. So we lived on Marta's parents futon for six months, but basically in order to. So we're hiring now and we'd be profitable within six months to a year. Basically. We were looking to hit 100k mrr within a year and that would bring us to. Well, I mean, will we be profitable?

And to answer your question directly, the beautiful thing is we don't actually have to acquire our consumers, so we go with the B2B enterprise model. So we actually use the user bases of a brand or retailer so they send their customers to our platform. So we don't actually have to acquire the end users.

Yeah, good point.

Hey, Will, would you sell stuff through this?

No, that's what I was going to ask. What do you do about trolls and bombers that could totally rip apart a product just because they're hating?

Yeah, so the way our data models work is that we find which users are quality and predictive, essentially because some people are just going to be saying Tinder, swiping on this thing and just saying yes to everything. And you can write data models that filter out for them. And that's kind of where my background in data science comes in. Wait, wait, wait.

They teach hater proofing and fixing physics classes now?

Well, they teach you how to find really distant signals in far away galaxies, which is kind of similar to finding like.

Does that have to do with Tinder? Again, sorry.

Anyway, so basically you just have to find the users that are actually quality and listen to them. And that's what our competitive advantage is, that we know how to do that.

How much of the back end is algorithm driven versus logic driven.

So it's neural nets, machine learning. So it's totally out there, 100%.

So 100%. So you can just grow as your data set grows and your user page grows, you don't have to sit there and program if this, then that there's no logic.

Exactly. And it learns which user.

And how do you deal with responses from consumers not just for the haters, but that don't make sense. So do you just parse and use NLP and just decide on your own if it's a valid response and discard everything else or what do you do?

So NLP is, I think, a little too early right now to really solve a lot of problems. So we mostly use a button interface and we kind of use a button interface and we're using preset choices.

So. But that's what I'm saying. So how do you. So for each product then you have a process where you define all the options as you think they should be. So you don't learn the options as they come based off the responses. You predefined them.

That's the future. But currently we send them down paths.

Depending on that explains it.

Okay, we have a company called Reply yes that you reply yes to buy something, but it actually knows if you say hell's yes or F. Yeah, it's kind of. Computers are smart like that.

So you said you're profitable, then why do you need the 150,000?

I didn't hear your excuse.

You mentioned that you're profitable. So why do you need this money? And you know, what are the user proceeds? How are you going to spend the money?

Yeah, so we're profitable in the sense that it's just the two of us and we make enough money to live and. But our biggest problem right now is actually account managing. These are big customers and you need to spend time with them. So I do product, Marta does sales, but we actually do account managing. So we need to hire an account manager, which we actually already have.

How have you gotten the sales that you've gotten to this point? Through your previous employers or previous connections.

So it's been a mix of cold emails, warm introductions, and how do you price it? So we price on a per product basis from 100 to $250 per product test.

Oh, man. Right, Right.

Woo.

Nice job. Bravo.

Good job, you guys. Good job. Good job.

Bravo. Bravo. Very, very good. All right, let's keep the momentum going.