How KERV.ai is Powering Contextual Video and Shoppable Ads with AI

Summary

Marketecture Podcast: Episode Summary

Title: How Curve AI is Powering Contextual Video and Shoppable Ads with AI
Host: Ari Paparo
Guest: Gary Mittman, Founder and CEO of Curve AI
Release Date: April 7, 2025

---

1. Introduction to Curve AI

In this episode of the Marketecture Podcast, host Ari Paparo engages in an insightful discussion with Gary Mittman, the Founder and CEO of Curve AI. The conversation delves into how Curve AI is revolutionizing the advertising landscape through advanced AI-driven video analysis and shoppable ad technologies.

2. Company Overview

Foundational Insights: Gary Mittman provides a foundational understanding of Curve AI, highlighting its eight-year journey in the industry.

• Location and Team:
  "We're operationally based in Austin, Texas, where our tech, media and operations and client service all live down there. Our marketing, finance and sales are predominantly focused in New York and I fly somewhere over Toledo on a regular basis." ([02:12])

• Company Size:
  "We're about 98 people at this point." ([02:35])

3. Core Technology and Offerings

AI-Driven Video Analysis and Shopability: Curve AI sits at the intersection of content and commerce, leveraging AI to analyze videos for contextual relevance and shoppability.

• Hybrid Product Approach:
  "It's a combination. It's a hybrid. Because the..." ([02:55])
  Gary explains that Curve AI combines creative and targeting functionalities to enhance ad placements within video content.

Practical Applications: A notable example includes Curve AI's collaboration with Warner Brothers, where they analyze entire content libraries to identify shoppable moments.

• "We did a pretty big project with them last year with a very large retailer where we analyze the show, correlate and match against the product catalogs to find what in the scene or in the scenes might be shoppable from that retailer." ([02:55] - [03:33])

4. Shoppable and Contextually Relevant Ads

Ad Integration: Curve AI integrates shoppable ads seamlessly into video content, ensuring they appear at optimal moments without disrupting the viewer experience.

• Dynamic Creative Generation:
  "So in that example, we do a dynamic creative, we build it on the fly." ([04:02])

Non-Publisher Use Cases: Gary addresses the challenge of publishers being hesitant to share context data, explaining Curve AI's dual-layer approach.

• "So the two layers are one is the shoppable ad units correlated to the content. The other one is contextually relevant ad placement." ([04:36])

5. Business Model and Partnerships

Revenue Streams: Curve AI operates on a dual revenue model, combining traditional CPM media strategies with technology-as-a-service offerings for enterprise clients.

• "In our ad business, it's traditional CPM media strategies. In our enterprise business... it's really technology as a service." ([08:50])

Strategic Partnerships: Recent integrations with platforms like Magnite, PubMatic, and Freewheel have expanded Curve AI's reach in programmatic advertising.

• "So to do the integration on the contextually relevant ad potting or shopability requires the integration with the platforms." ([10:14])

6. Technological Innovations and Patents

Proprietary AI Development: Gary elaborates on Curve AI's proprietary AI technology, which is safeguarded by two key patents.

• "Our first patent is about identifying the pixel edge or unique shape of an object... Our second patent is around analyzing the show, gathering the metadata..." ([07:22] - [07:52])

Live Content Application: Curve AI is currently applying its technology to live sports content, showcasing its capability to operate in real-time environments.

• "We're doing live sports right now." ([12:53] - [13:08])

7. Competitive Landscape and Challenges

Standing Against Giants: When questioned about competition from giants like Amazon, Google, and Facebook, Gary asserts Curve AI's unique position as a neutral party.

• "So we're a neutral party that sits in the middle." ([12:53])

Primary Challenges: The main challenges revolve around architectural relevance to shopability and navigating the complexities of existing publisher ecosystems.

• "The challenge is really the... current ecosystems, the struct publisher content, the contracts with talent as an example..." ([11:56] - [12:17])

8. Lightning Round

Towards the end of the episode, Ari engages Gary in a rapid-fire series of questions to uncover more about Curve AI's identity and operations.

• Number One Competitive Advantage:
  "Our contextual correlation of product catalog to shopability." ([11:49])

• Biggest Challenge:
  "Architectural relevance to shopability." ([11:56])

• Curve AI as an Animal:
  "A jaguar."
  "Moving fast and being efficient." ([13:15] - [13:19])

9. Conclusion

Ari wraps up the conversation by expressing appreciation for Gary Mittman's insights into Curve AI's innovative approach to contextual video and shoppable advertising. The episode offers a comprehensive look into how Curve AI is shaping the future of advertising through intelligent AI solutions.

---

Notable Quotes:

• "We're a neutral party that sits in the middle." — Gary Mittman ([12:53])
• "Our contextual correlation of product catalog to shopability." — Gary Mittman ([11:49])
• "A jaguar. Moving fast and being efficient." — Gary Mittman ([13:15])

---

For those interested in exploring more about Curve AI and their groundbreaking technologies, visit Curve AI's Website or tune into future episodes of the Marketecture Podcast for more industry-leading insights.

Transcript

[Ari]

This podcast is brought to you by Adelaide Media verification and measurement are undergoing major disruption. Legacy players are pivoting to performance. Advertising AI is reshaping brand safety and attention is replacing viewability. Adelaide is leading the shift with au, a new way to assess media quality that scores placements based on their potential to drive attention and outcomes. Before your ads run, think of it like a credit score for media. Finally, a clear view of quality. Before you buy, take the guesswork out of your investment strategy and try Adelaide AU on your next campaign. Welcome to marketecture, where you can get smart fast with in depth interviews of leading technology executives. I'm joined today by Gary Mittman. He is the co founder. Excuse me, the founder and CEO of Curve AI. So, Gary, thank you for being here.

[Gary Mittman]

My pleasure. Always a pleasure, sir.

[Ari]

I got the introduction right? More or less. Curve's been in the news. You've been making a lot of announcements and whatnot. But why don't we start at the basics? What is Curve? What does it do?

[Gary Mittman]

Who are these people?

[Ari]

Yeah.

[Gary Mittman]

So Curve's core essence is our AI for video analysis and contextual relevance and shopability. We're kind of at the vortex between content and commerce. We sit right in the middle. We've been doing this for eight years, so all the new commerce media stuff to us is entertaining. We've been doing it for a while.

[Ari]

Yeah, Vertex, not Vortex. Vortex would be like a whirlwind. Vertex is the intersection. But we know what you mean, correct?

[Gary Mittman]

Vertex. I'll go with that. So the core of our system is an AI platform that analyzes both ads and content and makes it either shoppable or contextually relevant. Our customers include the holdcos, all the primary Holdcos for shoppable ad units and distribution through Programmatic, along with a lot of the major content owners for analysis for contextual relevance and shopability.

[Ari]

Cool. So tell me about the company. How big is it? Where is it based? How long have you been doing it?

[Gary Mittman]

So I started the company eight years ago. We're operationally based in Austin, Texas, where our tech, media and operations and client service all live down there. Our marketing, finance and sales are predominantly focused in New York and I fly somewhere over Toledo on a regular basis.

[Ari]

And how big is the company?

[Gary Mittman]

We're about 98 people at this point.

[Ari]

It's a good size. Okay, so let's talk about this. So you analyze videos. You understand what they're about? I imagine contextually. So how does the. How does the product use that? Is it. Is it really a creative product or Is it a targeting product or some combination?

[Gary Mittman]

It's a combination. It's a hybrid. Because the. So I'll give you some practical examples. So Warner Brothers, we analyze their entire library and we contextualize it for both ad potting and or shopability. We did a pretty big project with them last year with a very large retailer where we analyze the show, correlate and match against the product catalogs to find what in the scene or in the scenes might be shoppable from that retailer. We then create an ad slot right at the right scene, at the right moment to put a shoppable ad unit there.

[Ari]

Right. Is that an overlay unit or is it linear?

[Gary Mittman]

It is a stitched in. Traditional ad.

[Ari]

Traditional ad. Right. Okay. So that use case is interesting. So the character is drinking orange juice and you throw a tropical and add in something like that.

[Gary Mittman]

Well, a little more complex than that, but yeah, that's the general idea.

[Ari]

So that example is a publisher served ad. Right. It's in the ad server. What about the creative? Is it a custom creative or is it just a typical creative?

[Gary Mittman]

So in that example, we do a dynamic creative, we build it on the fly. So once we have the catalog ingested, we have the scene analyzed. We can then dynamically compile of AZ tag that gets pushed out.

[Ari]

Interesting. Okay, got it. And then what's the use case in sort of a non publisher use case? Because one of the challenges with contextual video is that the publishers don't really want the buyers to know the context. They're very shy about giving any data about the videos. So how does it work in your system where buyers get to choose?

[Gary Mittman]

So the two layers are one is the shoppable ad units correlated to the content. The other one is contextually relevant ad placement. Right. So the contextually relevant ad placement is one that's currently being sold direct from the publisher themselves. We are looking now at advancing it into programmatic, but that's the next step.

[Ari]

Okay, so most of the business is publisher centric currently.

[Gary Mittman]

On the content side.

[Ari]

Right.

[Gary Mittman]

We do have our ad unit side that we. As an example, Disney sells our interactive shoppable ads across their platform. Oh.

[Ari]

So explain how that works where there's not one specific advertiser. Or is it. It's still a direct relationship.

[Gary Mittman]

So yeah. So a client calls up, Disney wants to run a campaign and they say, I want it shoppable. Disney turns to curve and we make the asset interactive, hand it back and they run it.

[Ari]

I see. So is it possible to use that in a sort of DSP or DCO.

[Gary Mittman]

100%. It's a tag.

[Ari]

It's a tag.

[Gary Mittman]

It can run programmatically.

[Ari]

Right. Okay. You just may not have the context from the publisher if you're buying it programmatically.

[Gary Mittman]

Correct. And if they're selling it directly, they're selling it based on their placement criteria.

[Ari]

Yeah. Tell me about the creative. So is this templated? You have AI in your name. So I want to know, are you like using the magical, magical cloud intelligence to create these ads?

[Gary Mittman]

No, actually, it's. Well, there's a couple things. So the AI is used to correlate and match objects to catalogs to products. Right. So we, we, we're working right now as an example with the. I can't name names yet because I don't. It's not public, but a large publisher in real time. Live analysis. So we're analyzing sports content live and then correlating and matching for ad placement and or shopability. The AI is powering that. So when we look at content and let's say the Warner Brothers example, the AI platform analyzes the show, identifies the objects, looks at the product catalog, correlates and match in finding relevant, exact or not, depending on the. When we match, we match by percentage ratio. So an exact match, a 50%, whatever it might be. So we can identify and correlate that. When we do that, then we make the ad. So the AI powers video analysis, ad analysis, content correlation to product catalog. We also, by the way, use the AI. We work with a couple of companies where we're doing ad compliance, where we're doing the ad approval, in other words, analyzing the ads for nudity, profanity, all the things that they're manually doing as required. Now we're doing it with the AI.

[Ari]

So how did you build this AI technology? I know you mentioned something earlier about a patent before we got on the call.

[Gary Mittman]

Yeah. We actually have two patents. Our first patent is about identifying the pixel edge or unique shape of an object, the unique curved shape of an object so that we can identify it in content and then correlate a match. And then for linking out for purchase is the patent. The second patent is around analyzing the show, gathering the metadata, and how we use that to correlate and match against product catalogs.

[Ari]

And was that all built internally?

[Gary Mittman]

Yeah, all internally. This is all proprietary, right?

[Ari]

And does this stuff work? Do you have a case study showing performance?

[Gary Mittman]

No, it doesn't work.

[Ari]

First person in the history of architecture to admit it. It just looks good. It's good on the slide.

[Gary Mittman]

It's great. It's a great slide. No, of course it works. We have incredible case studies. We've got. Well, I'll go back to Warner Brothers. We did a great campaign with them with Wayfair. It's public knowledge, so I can say that that was hugely successful. Wayfair came back loving it. On our shoppable ad units, we've got endless case studies with partners throughout the ecosystem. We're partnered now with resellers like Disney, the OEMs, and a lot of different people who are reselling the shoppable ad units.

[Ari]

Okay, cool. And tell me about the business model. Is it cpm? How's it working with publishers? Setup, costs, et cetera?

[Gary Mittman]

So there's really two lines. So in our ad business, it's traditional CPM media strategies. In our enterprise business, where we do the correlation, contextual relevance, et cetera, it kind of varies depending on the publisher nuances. Each of the publishers have different criteria and variables that they require, current structures and various things that we have to work within. So the structures are a little different. It's really technology as a service for the majority of those people where we provide them the capabilities of what we do. The best scenario is we host the data for them. We become their dmp, in essence, on the contextual relevant metadata to be utilized in the multiple ways.

[Ari]

Right. And what's the entry point like? Is it. Is it expensive service, you have a lot of training, etc. Or. Or do you do signal campaigns?

[Gary Mittman]

We try to do all the work. So our intention is to make it as easy as possible for the clients. We've got a full team doing really all the labor. In essence, the simpler we make it for the customer, the better.

[Ari]

Yeah, got it. Now, recently you've been in the news because you announced partnerships, I think, with Magnite, Pugmatic and Freewheel over the past year, more or less. Can you tell us what does that mean? Who's the customer and how does it actually work?

[Gary Mittman]

Sure. So to do the integration on the contextually relevant ad potting or shopability requires the integration with the platforms. So Freewheel, as an example, we work with them, with Warner Brothers and a few other people to manage the contextual ad slotting and the placement and the timing of pushing out relevant ads at the right place at the right time.

[Ari]

Right. So in the free will case, it's an ad server, so they need to figure out where to insert the ads and you're giving them some signal of some kind.

[Gary Mittman]

Correct, Right, correct. And that signal then can be Pushed out to the pubmatics of the world so that we can then have programmatic opportunity for purchasing in that slot.

[Ari]

And when you do that, do you just pass them as sort of like key values of some kind?

[Gary Mittman]

Yeah.

[Ari]

Do you use like the IB Standard taxonomy or something video specific?

[Gary Mittman]

It's an interesting word, taxonomy. We've been laughing about that recently.

[Ari]

So laughing about taxonomy, we have something in common.

[Gary Mittman]

I think that your scope 3 description actually was pretty good around this. In that taxonomy is really for humans.

[Ari]

Yeah, it is. Taxonomies are for humans.

[Gary Mittman]

If a human being is doing buying and needs to categorize it in a way they understand, that's a taxonomy. But if you're looking at AI, it doesn't need that. It has a deeper understanding. So in the short term, yes. In the long term, no. An answer to your question.

[Ari]

Okay, that sounds good. Let's. Let's move to a lightning round. So, relatively quick questions, quick answers. What's your number one competitive advantage?

[Gary Mittman]

Our contextual correlation of product catalog to shopability.

[Ari]

What's your biggest challenge?

[Gary Mittman]

Architectural relevance to shopability.

[Ari]

Do you want to expand on that a little bit? Why? It's a challenge.

[Gary Mittman]

No, no, the challenge is really the. The current ecosystems, the struct publisher content, the contracts with talent as an example, all the different variables within that are, I think, the challenges.

[Ari]

Okay, why won't companies like Amazon, Google, Facebook just ultimately crush you?

[Gary Mittman]

Well, Amazon's already doing this.

[Ari]

They're already crushing.

[Gary Mittman]

No, no, they're doing these strategies internally. We actually met with them five years ago, and all of a sudden they're doing exactly with our presentation. Kind of pisses me off.

[Ari]

Well, they got the product catalog and they have a lot of videos, so.

[Gary Mittman]

Yeah, I figured it out pretty quickly. But I think the key is that I think the content owners, the major publishers, don't want to be giving their data away to the Amazons and the Googles.

[Ari]

Yeah.

[Gary Mittman]

So we're a neutral party that sits in the middle.

[Ari]

Hey, this is a little off topic. Not in a lightning round, but it just reminded me, is there any applicability here for live content?

[Gary Mittman]

We're doing live content now, so you.

[Ari]

Can scan it fast enough.

[Gary Mittman]

We're doing live sports right now.

[Ari]

All right, cool. Okay, last question. If Curve was an animal, what animal would it be?

[Gary Mittman]

A jaguar.

[Ari]

Why?

[Gary Mittman]

Moving fast and being efficient.

[Ari]

All right, I'll take it. As long as it's not a lion. No more lions. All right. Terry Mittman, the founder of Curve AI, thank you for being here.

[Gary Mittman]

My pleasure, Ari. Thank you. And it's great to meet you.

[Ari]

Great, Great talking. Thank you for listening to the marketecture podcast. New episodes come out every Friday and an insightful vendor interview is published each Monday. You can subscribe to our library of hundreds of executive interviews at marketecture tv. You can also sign up for free for our weekly newsletter with my original strategic insights on the week's news at News Market. And if you're feeling social, we operate a vibrant Slack community that you can apply to join at adtechgod. Com.