How 3 AI Startups Are Rewriting Marketing - BranchLab, SWYM, Veylan
Jeremy Kagan · June 23, 2025
ai in marketing
In this session of AI Insiders, presenters from Branch Lab, SWYM, and Veilan explain how AI-native platforms are reshaping marketing. They discuss advancements from healthcare audience targeting and programmatic media optimization to an integrated AI operating system for digital advertising, emphasizing efficiency, data privacy, and transformative workflow improvements.
0:05 [David]: Hey everyone. Welcome to another edition of AI Insiders by AI Marketers Guild. I'm David Berrkwitz, and I'm excited to host my friend, investor, and professor Jeremy Kagan from Textbook Ventures. Jeremy, welcome—I’m excited to hear about your portfolio and to share this conversation with some great guests and insightful questions.
1:09 [Jeremy]: I’m very excited to be here. It’s great to see you, David, and to rejoin the AI Marketers Guild after many years at Market. It’s wonderful to see old friends together. If you’re listening to this recording, you missed the dad jokes—we’ll keep them on record. I now have a production team at Market, so that part is behind me.
2:08 [Jeremy]: I’m thrilled to discuss the companies that illustrate our points today. As a professor and venture capitalist with Textbook Ventures, which invests in talent from schools like Columbia, Cornell Tech, and NYU, I get an early look at innovative companies. AI in marketing is one of the hottest areas, consistently improving performance—even among those who normally underperform.
3:09 [Jeremy]: McKinsey finds that corporations using AI tools generally outperform those who don’t, with even the lower achievers making significant gains. The takeaway is simple: if more people work with AI, everyone benefits. My advice to new graduates is to explore AI tools now—if you don’t, someone who masters them might take your job.
4:11 [Jeremy]: Many equate ChatGPT with AI; while it’s powerful, it’s just one example of generative AI, which creates content much like an advanced autocorrect. Just as autocorrect sometimes makes errors, these tools can also generate inaccurate or “hallucinated” responses. Essentially, AI analyzes vast quantities of text data to predict relationships between words.
5:11 [Jeremy]: A professor once told me that if you ask someone for a number between 1 and 5, each number should have about a 20% chance of being picked. However, large language models sometimes favor certain responses, meaning they’re less mathematically precise. We don’t need to understand all the underlying math to effectively use these tools—much like driving a car without knowing how its engine works.
6:12 [Jeremy]: AI leverages context by examining the words around each term, which is why it can provide accurate responses when given more data. This overview is crude, but it sets the stage for discussing real-world applications. Would anyone prefer more on the mechanics, or should I skip to how AI is used in marketing?
8:14 [Jeremy]: Beyond generating content to overcome blank-page syndrome, AI finds patterns for targeting, bidding, and campaign evaluation. One key question arises: if AI automates tasks once performed by experienced professionals, do we risk losing valuable expertise? I believe we should embrace these tools—using them is comparable to employing a calculator to split a dinner bill.
10:18 [Jeremy]: The difference between generative and applied AI isn’t an either/or matter—generative is simply one form of AI. Now, I’ll introduce our first company. Our first presenter is from Branch Lab; please welcome Josh.
11:20 [Josh – Branch Lab]: Thanks, Jeremy. It’s a pleasure to be here. I recently became a dad for the second time, so I’m a bit sleep deprived, but let’s dive in. I’m sharing my screen now. At Branch Lab, we use AI to solve very targeted problems, making bold predictions about the future—for instance, predicting that most websites will eventually be replaced by AI agents creating content in dynamic, interactive formats. This shift is evident even at Google, where traditional search ad revenue is being eroded by AI responses.
12:22 [Josh – Branch Lab]: For future advertising technology, we see AI-native systems dominating—from strategic and creative agents to analytics, optimization, brand safety, media, and audience targeting. In healthcare marketing, the approach shifts from simply targeting diagnosed individuals to predicting outcomes, such as increasing prescription fills by identifying those likely to benefit from a drug.
13:26 [Josh – Branch Lab]: Our software interface resembles an AI agent. The user states an outcome—like boosting prescription fills—without specifying conventional targeting details. Behind the scenes, we analyze healthcare journeys from extensive data to predict diagnoses and deliver tailored messaging, all while preserving consumer privacy.
14:29 [Josh – Branch Lab]: The system constructs audiences using nonsensitive, aggregate data, which can be ported to social platforms, connected TV, or even future AI agents. Instead of relying on traditional identifiers like cookies, we use probabilistic models that ensure data portability and privacy.
15:33 [Josh – Branch Lab]: The front end works like platforms such as SATBT or Perplexity, asking users for desired outcomes and then constructing patient journeys and audience segments based on predicted events rather than known diagnoses.
16:34 [Josh – Branch Lab]: We analyze patient populations at key moments, recommending messaging strategies that target both consumers and healthcare professionals for informed conversations. Campaign efficacy is measured by correlating outcomes like prescription fills with our predicted audience data.
17:40 [Josh – Branch Lab]: Branch Lab’s curated audiences are live across channels—from addressable TV to Meta. I’ll now take a moment for questions, especially regarding data privacy and sources.
23:56 [Audience]: How do you target when health data is anonymized and individual identities aren’t known?
25:00 [Josh – Branch Lab]: In our context, anonymized means that names and exact addresses are removed, while key details—such as age, gender, generalized location (like a zip code segment), facility information, and prescribing physician—remain. We then augment this with additional demographic data to build profiles that predict health events without revealing personal identities.
27:07 [Josh – Branch Lab]: Essentially, the system outputs a probability-based audience model that is both portable and privacy-preserving. We’re proud to work with half of the world’s top ten pharmaceutical manufacturers, and recent Harvard Law research even cites Branch Lab as a privacy-forward solution.
29:15 [David]: Let’s now move on to our next presenter—Andy from SWYM. We’ll circulate relevant links and contact details via chat.
30:21 [Andy – SWYM]: Thanks, David. I’m one of the co-founders of SWYM. We launched around 18–20 months ago, and today we work with over 35 agencies and 50 advertisers to optimize programmatic media spend. In digital marketing, about 25–35% of the $700 billion spent annually is wasted on low-quality ad placements—the so-called “lemon market.”
31:27 [Andy – SWYM]: A “lemon market” occurs when there’s significant information asymmetry between buyers and sellers, leading advertisers to buy low-quality ad placements due to issues like mis-targeting, fraud, or non-viewability. While many optimize the buy side, we’ve discovered that the supply side is largely unaware of performance quality.
32:30 [Andy – SWYM]: In programmatic advertising, publishers and sell-side platforms lack insight into what makes an impression high-performing, so agencies end up spending their entire budgets on billions of bid requests that include many lemons.
33:33 [Andy – SWYM]: To address this, we built an AI-driven algorithmic approach for supply curation. We integrate with major supply-side platforms (including Google AdX, Magnite, OpenX, PubMatic, and others) to access nearly 100% of the open web while remaining agnostic on the buy side.
34:33 [Andy – SWYM]: Our system establishes a learning feedback loop based on in-market performance. By analyzing which ad placements meet critical KPIs—be it lower cost per acquisition or higher engagement—we continuously adjust our supply curation strategy.
35:34 [Andy – SWYM]: Imagine walking into a grocery store where you must spend every cent, yet most items on the shelves are low quality. In this scenario, you’d end up with a basket full of lemons. Our AI system filters the bidstream, replacing lemons with “cherries” that are high-performing and tailored to each campaign.
36:34 [Andy – SWYM]: Our algorithm learns from daily performance data—such as ad dimensions, website context, geography, time-of-day, and device type—and adapts bidding strategies in real time by knowing when to bid higher or lower based on these signals.
37:40 [Andy – SWYM]: This approach has simplified the supply chain dramatically. For example, a financial services client saw a 36% reduction in cost per acquisition; a connected TV campaign delivered a 23% reduction in cost per completed view; and an Amazon DSP client experienced consistent outperformance—all achieved by filtering for effective supply.
38:40 [Andy – SWYM]: Our solution is tech-agnostic, channel-rich, easy to activate and measure, and best of all, it costs advertisers nothing—the sell side funds our efforts. We’re essentially changing the media buying landscape to reduce waste and drive better campaign outcomes.
41:46 [Andy – SWYM]: I’ll open the floor for any last questions on our platform before we hand the stage over to our third presenter, from Veilan.
45:53 [Daniel – Veilan]: Thanks, Jeremy. I’m Daniel Mian, co-founder of Veilan, an AI-native operating system for advertising. Our platform is built on technology developed over the past 10 years—originally in hedge funds, defense, and cybersecurity—and addresses inefficiencies where legacy systems operate in silos by connecting and consolidating them.
46:52 [Daniel – Veilan]: Veilan integrates with your existing infrastructure—whether it’s AWS, Google Cloud, databases, email systems, or ad servers—so that disparate systems speak the same language. Our platform listens to your brand data and accelerates the process from ideation to execution.
47:59 [Daniel – Veilan]: Digital advertising is a massive industry, with almost $800 billion spent globally. Yet inefficiencies in the system mean that much of that spend goes to waste. Veilan transforms these processes by providing a conversational interface that consolidates all the past campaign data and insights.
49:01 [Daniel – Veilan]: For example, generating a proposal used to involve combing through legacy systems and manual data searches—a process that could take days. With Veilan, the process is streamlined and produces on-brand proposals within minutes using historical performance insights.
49:58 [Daniel – Veilan]: Our platform supports everything from brainstorming and storyboarding to campaign planning, creative automation, and ad distribution. It can generate digital ad formats on the fly—whether that's dynamic rich media, static banners, or social videos—by leveraging preapproved creative assets.
50:58 [Daniel – Veilan]: Once a campaign is live, our system continuously ingests data from both demand and supply sides, enabling real-time optimization. This feedback loop allows the platform to intelligently adjust targeting and creative strategies as the campaign progresses.
51:58 [Daniel – Veilan]: Veilan dramatically reduces production times, providing micro insights from past campaigns that would otherwise be impossible to derive manually. It connects the entire advertising pipeline—from strategy through execution—in a seamless, AI-powered workflow.
53:06 [Daniel – Veilan]: For instance, a CEO can quickly analyze past ad performance, determine optimal calls-to-action, and see which formats perform best—empowering teams to break down legacy silos and work more efficiently.
54:06 [Daniel – Veilan]: In essence, Veilan replaces labor-intensive manual processes with automated intelligence that enhances creativity and performance. This isn’t about replacing humans—the transformation is in how teams use AI to expand what they can achieve.
55:14 [Daniel – Veilan]: The platform continuously optimizes targeting, creative messaging, and campaign distribution by learning in real time. It delivers actionable insights so brands can achieve better outcomes faster.
56:16 [Daniel – Veilan]: Our AI fine-tunes campaigns on the fly, ensuring that every aspect—from creative to distribution—is adjusted according to the latest performance data. The result is a significant leap in efficiency for digital advertising workflows.
57:20 [Daniel – Veilan]: Ultimately, Veilan empowers teams by aggregating vast amounts of data and generating insights that help transform campaign planning and execution. The shift isn’t just technological—it’s a complete transformation in how advertising is strategized and delivered.
58:24 [Daniel – Veilan]: AI is not simply a tool; it shifts power within organizations. Embracing these advanced tools is essential for continued innovation in marketing. I appreciate your attention and will now conclude my presentation.
59:21 [David]: Thank you to all our presenters—Josh, Andy, and Daniel—for an insightful session on how Branch Lab, SWYM, and Veilan are rewriting marketing with AI. Feel free to stay on for questions, share contact information, and continue the conversation. Thanks, everyone, for joining us today!
0:05 [David]: Hey everyone. Welcome to another edition of AI Insiders by AI Marketers Guild. I'm David Berrkwitz, and I'm excited to host my friend, investor, and professor Jeremy Kagan from Textbook Ventures. Jeremy, welcome—I’m excited to hear about your portfolio and to share this conversation with some great guests and insightful questions.
1:09 [Jeremy]: I’m very excited to be here. It’s great to see you, David, and to rejoin the AI Marketers Guild after many years at Market. It’s wonderful to see old friends together. If you’re listening to this recording, you missed the dad jokes—we’ll keep them on record. I now have a production team at Market, so that part is behind me.
2:08 [Jeremy]: I’m thrilled to discuss the companies that illustrate our points today. As a professor and venture capitalist with Textbook Ventures, which invests in talent from schools like Columbia, Cornell Tech, and NYU, I get an early look at innovative companies. AI in marketing is one of the hottest areas, consistently improving performance—even among those who normally underperform.
3:09 [Jeremy]: McKinsey finds that corporations using AI tools generally outperform those who don’t, with even the lower achievers making significant gains. The takeaway is simple: if more people work with AI, everyone benefits. My advice to new graduates is to explore AI tools now—if you don’t, someone who masters them might take your job.
4:11 [Jeremy]: Many equate ChatGPT with AI; while it’s powerful, it’s just one example of generative AI, which creates content much like an advanced autocorrect. Just as autocorrect sometimes makes errors, these tools can also generate inaccurate or “hallucinated” responses. Essentially, AI analyzes vast quantities of text data to predict relationships between words.
5:11 [Jeremy]: A professor once told me that if you ask someone for a number between 1 and 5, each number should have about a 20% chance of being picked. However, large language models sometimes favor certain responses, meaning they’re less mathematically precise. We don’t need to understand all the underlying math to effectively use these tools—much like driving a car without knowing how its engine works.
6:12 [Jeremy]: AI leverages context by examining the words around each term, which is why it can provide accurate responses when given more data. This overview is crude, but it sets the stage for discussing real-world applications. Would anyone prefer more on the mechanics, or should I skip to how AI is used in marketing?
8:14 [Jeremy]: Beyond generating content to overcome blank-page syndrome, AI finds patterns for targeting, bidding, and campaign evaluation. One key question arises: if AI automates tasks once performed by experienced professionals, do we risk losing valuable expertise? I believe we should embrace these tools—using them is comparable to employing a calculator to split a dinner bill.
10:18 [Jeremy]: The difference between generative and applied AI isn’t an either/or matter—generative is simply one form of AI. Now, I’ll introduce our first company. Our first presenter is from Branch Lab; please welcome Josh.
11:20 [Josh – Branch Lab]: Thanks, Jeremy. It’s a pleasure to be here. I recently became a dad for the second time, so I’m a bit sleep deprived, but let’s dive in. I’m sharing my screen now. At Branch Lab, we use AI to solve very targeted problems, making bold predictions about the future—for instance, predicting that most websites will eventually be replaced by AI agents creating content in dynamic, interactive formats. This shift is evident even at Google, where traditional search ad revenue is being eroded by AI responses.
12:22 [Josh – Branch Lab]: For future advertising technology, we see AI-native systems dominating—from strategic and creative agents to analytics, optimization, brand safety, media, and audience targeting. In healthcare marketing, the approach shifts from simply targeting diagnosed individuals to predicting outcomes, such as increasing prescription fills by identifying those likely to benefit from a drug.
13:26 [Josh – Branch Lab]: Our software interface resembles an AI agent. The user states an outcome—like boosting prescription fills—without specifying conventional targeting details. Behind the scenes, we analyze healthcare journeys from extensive data to predict diagnoses and deliver tailored messaging, all while preserving consumer privacy.
14:29 [Josh – Branch Lab]: The system constructs audiences using nonsensitive, aggregate data, which can be ported to social platforms, connected TV, or even future AI agents. Instead of relying on traditional identifiers like cookies, we use probabilistic models that ensure data portability and privacy.
15:33 [Josh – Branch Lab]: The front end works like platforms such as SATBT or Perplexity, asking users for desired outcomes and then constructing patient journeys and audience segments based on predicted events rather than known diagnoses.
16:34 [Josh – Branch Lab]: We analyze patient populations at key moments, recommending messaging strategies that target both consumers and healthcare professionals for informed conversations. Campaign efficacy is measured by correlating outcomes like prescription fills with our predicted audience data.
17:40 [Josh – Branch Lab]: Branch Lab’s curated audiences are live across channels—from addressable TV to Meta. I’ll now take a moment for questions, especially regarding data privacy and sources.
23:56 [Audience]: How do you target when health data is anonymized and individual identities aren’t known?
25:00 [Josh – Branch Lab]: In our context, anonymized means that names and exact addresses are removed, while key details—such as age, gender, generalized location (like a zip code segment), facility information, and prescribing physician—remain. We then augment this with additional demographic data to build profiles that predict health events without revealing personal identities.
27:07 [Josh – Branch Lab]: Essentially, the system outputs a probability-based audience model that is both portable and privacy-preserving. We’re proud to work with half of the world’s top ten pharmaceutical manufacturers, and recent Harvard Law research even cites Branch Lab as a privacy-forward solution.
29:15 [David]: Let’s now move on to our next presenter—Andy from SWYM. We’ll circulate relevant links and contact details via chat.
30:21 [Andy – SWYM]: Thanks, David. I’m one of the co-founders of SWYM. We launched around 18–20 months ago, and today we work with over 35 agencies and 50 advertisers to optimize programmatic media spend. In digital marketing, about 25–35% of the $700 billion spent annually is wasted on low-quality ad placements—the so-called “lemon market.”
31:27 [Andy – SWYM]: A “lemon market” occurs when there’s significant information asymmetry between buyers and sellers, leading advertisers to buy low-quality ad placements due to issues like mis-targeting, fraud, or non-viewability. While many optimize the buy side, we’ve discovered that the supply side is largely unaware of performance quality.
32:30 [Andy – SWYM]: In programmatic advertising, publishers and sell-side platforms lack insight into what makes an impression high-performing, so agencies end up spending their entire budgets on billions of bid requests that include many lemons.
33:33 [Andy – SWYM]: To address this, we built an AI-driven algorithmic approach for supply curation. We integrate with major supply-side platforms (including Google AdX, Magnite, OpenX, PubMatic, and others) to access nearly 100% of the open web while remaining agnostic on the buy side.
34:33 [Andy – SWYM]: Our system establishes a learning feedback loop based on in-market performance. By analyzing which ad placements meet critical KPIs—be it lower cost per acquisition or higher engagement—we continuously adjust our supply curation strategy.
35:34 [Andy – SWYM]: Imagine walking into a grocery store where you must spend every cent, yet most items on the shelves are low quality. In this scenario, you’d end up with a basket full of lemons. Our AI system filters the bidstream, replacing lemons with “cherries” that are high-performing and tailored to each campaign.
36:34 [Andy – SWYM]: Our algorithm learns from daily performance data—such as ad dimensions, website context, geography, time-of-day, and device type—and adapts bidding strategies in real time by knowing when to bid higher or lower based on these signals.
37:40 [Andy – SWYM]: This approach has simplified the supply chain dramatically. For example, a financial services client saw a 36% reduction in cost per acquisition; a connected TV campaign delivered a 23% reduction in cost per completed view; and an Amazon DSP client experienced consistent outperformance—all achieved by filtering for effective supply.
38:40 [Andy – SWYM]: Our solution is tech-agnostic, channel-rich, easy to activate and measure, and best of all, it costs advertisers nothing—the sell side funds our efforts. We’re essentially changing the media buying landscape to reduce waste and drive better campaign outcomes.
41:46 [Andy – SWYM]: I’ll open the floor for any last questions on our platform before we hand the stage over to our third presenter, from Veilan.
45:53 [Daniel – Veilan]: Thanks, Jeremy. I’m Daniel Mian, co-founder of Veilan, an AI-native operating system for advertising. Our platform is built on technology developed over the past 10 years—originally in hedge funds, defense, and cybersecurity—and addresses inefficiencies where legacy systems operate in silos by connecting and consolidating them.
46:52 [Daniel – Veilan]: Veilan integrates with your existing infrastructure—whether it’s AWS, Google Cloud, databases, email systems, or ad servers—so that disparate systems speak the same language. Our platform listens to your brand data and accelerates the process from ideation to execution.
47:59 [Daniel – Veilan]: Digital advertising is a massive industry, with almost $800 billion spent globally. Yet inefficiencies in the system mean that much of that spend goes to waste. Veilan transforms these processes by providing a conversational interface that consolidates all the past campaign data and insights.
49:01 [Daniel – Veilan]: For example, generating a proposal used to involve combing through legacy systems and manual data searches—a process that could take days. With Veilan, the process is streamlined and produces on-brand proposals within minutes using historical performance insights.
49:58 [Daniel – Veilan]: Our platform supports everything from brainstorming and storyboarding to campaign planning, creative automation, and ad distribution. It can generate digital ad formats on the fly—whether that's dynamic rich media, static banners, or social videos—by leveraging preapproved creative assets.
50:58 [Daniel – Veilan]: Once a campaign is live, our system continuously ingests data from both demand and supply sides, enabling real-time optimization. This feedback loop allows the platform to intelligently adjust targeting and creative strategies as the campaign progresses.
51:58 [Daniel – Veilan]: Veilan dramatically reduces production times, providing micro insights from past campaigns that would otherwise be impossible to derive manually. It connects the entire advertising pipeline—from strategy through execution—in a seamless, AI-powered workflow.
53:06 [Daniel – Veilan]: For instance, a CEO can quickly analyze past ad performance, determine optimal calls-to-action, and see which formats perform best—empowering teams to break down legacy silos and work more efficiently.
54:06 [Daniel – Veilan]: In essence, Veilan replaces labor-intensive manual processes with automated intelligence that enhances creativity and performance. This isn’t about replacing humans—the transformation is in how teams use AI to expand what they can achieve.
55:14 [Daniel – Veilan]: The platform continuously optimizes targeting, creative messaging, and campaign distribution by learning in real time. It delivers actionable insights so brands can achieve better outcomes faster.
56:16 [Daniel – Veilan]: Our AI fine-tunes campaigns on the fly, ensuring that every aspect—from creative to distribution—is adjusted according to the latest performance data. The result is a significant leap in efficiency for digital advertising workflows.
57:20 [Daniel – Veilan]: Ultimately, Veilan empowers teams by aggregating vast amounts of data and generating insights that help transform campaign planning and execution. The shift isn’t just technological—it’s a complete transformation in how advertising is strategized and delivered.
58:24 [Daniel – Veilan]: AI is not simply a tool; it shifts power within organizations. Embracing these advanced tools is essential for continued innovation in marketing. I appreciate your attention and will now conclude my presentation.
59:21 [David]: Thank you to all our presenters—Josh, Andy, and Daniel—for an insightful session on how Branch Lab, SWYM, and Veilan are rewriting marketing with AI. Feel free to stay on for questions, share contact information, and continue the conversation. Thanks, everyone, for joining us today!