Faster Smarter Scalable Creative Inside BCMs Ventas AI-Powered Workflow

In this session, David Berkowitz and Max Cammarota from BCM discuss the Ventas AI-powered workflow that scales creative production for performance marketing.

They detail how advanced AI tools and a structured creative process can generate diverse ads, optimize performance, and accommodate platform-specific requirements while ensuring brand consistency.

0:05 – David:
"Hello everyone, I'm David Berkowitz. Welcome back to another edition of AI Insiders by AI Marketers Guild. I'm excited to be here live with the team from Marketers Guild and our featured speaker today. Max Cammarota from BCM has long been an active supporter and advocate of our community. I've learned so much from him—even from our first meeting two years ago. Today, I’m thrilled to have him share his insights with us."

1:10 – David:
"John introduced us, and I've learned a lot from both him and Max. Max, you were on a panel with me in April, so please share what you’re doing. I’m excited for everyone in the community to benefit from today’s discussion. Welcome, Max."

2:16 – Max:
"Thanks for having me. We appreciate the partnership and the work we’ve done together over the past months and years. Can you see my screen? Here we go. Today, I'll talk about Ventas AI—a creative workflow powered by AI."

3:26 – Max:
"Ventas AI is an AI-powered workflow to develop creative. We leverage the latest AI technology to enhance our creative process, making it faster, smarter, and more efficient. With it, we create high volumes of diverse ads that fuel platform algorithms. Business growth requires expanding channels, and each channel’s algorithm performs better with diverse creative. Traditional production couldn’t keep up, and that’s why Ventas AI was born."

4:28 – Max:
"Here are some results: We've launched thousands of ads since 2022 with a 100% client retention rate, won the Newsweek AI Impact Award for marketing, reduced cost per lead by 66%, achieved ROAS hikes above 150%, and boosted business revenue by 15% year-over-year."

5:32 – Max:
"In 2022, we faced challenges as paid search became competitive and expensive. We needed to expand channels to reach new users effectively, but producing a high volume of varied creative at scale was costly. This challenge led to the creation of Ventas AI. I remember presenting these challenges to our partners at BCM in Stamford, Connecticut. I pitched a creative process that later evolved into Ventas AI."

6:33 – Max:
"The algorithms of these platforms are evolving quickly. User activity triggers the system to retrieve relevant ads from a vast corpus of creative materials. Instead of a prescriptive AB testing approach, we now create many ad variants through multivariate testing. This allows the algorithm to distribute spend effectively and improve overall performance."

7:39 – Max:
"In our previous strategy, we prescribed targeting and relied on AB testing for the best ad. Today, we feed algorithms a diverse set of messages, allowing them to determine the best performing combinations. This leads to continuous improvement in both middle-funnel metrics, like clickthrough rates, and conversion rates."

8:47 – Max:
"The creative bottleneck was always an issue because traditional production couldn’t meet the scale needed. We designed Ventas AI to accelerate market entry, continuously refresh creative, and test new ideas quickly. Speed is key; after identifying a winning ad, waiting 8 to 12 weeks for production wasn’t feasible."

9:47 – Max (responding to a question from Mike):
"Our platform works by pulling individual elements—post copy, images, designs—in isolation. We then combine them into various formats. For regulated industries like healthcare, we incorporate human checkpoints to ensure all produced creative is preapproved and brand compliant."

10:51 – Max:
"In terms of generating posts for various social platforms, Ventas AI handles both situations. We use generative AI to create copy and imagery and also manipulate existing client assets to fit our creative formats."

11:53 – Max:
"Approvals happen in isolation to avoid production bottlenecks. We also utilize final ad banks to give our clients transparent views into what is going live, ensuring quality at scale while maintaining speed."

12:59 – Max:
"Brand voice is maintained by training our system with brand guidelines. Human checkpoints ensure that creative output remains on brand. If generative outputs do not meet the brand standards, we seek alternative assets, including client-supplied or stock imagery."

14:02 – David (interjecting a question from Mike):
"Are you configuring the system using approved content libraries, or do you iterate based on client examples?"

15:09 – Max:
"It’s two-part. The platform pulls approved elements and recombines them into ad variants. We work closely with clients, especially in highly regulated fields, to ensure all messaging is preapproved and compliant."

16:12 – Max:
"Our system is flexible. For each medium—whether static posts, video, or other formats—we either create outputs using generative AI or adapt existing assets as needed. This provides versatility across platforms."

17:13 – Max:
"Regarding final content approvals, while there is a human approval process for isolated elements, final ad approval is expedited since many individual elements have already been preapproved."

18:19 – Max:
"As for maintaining brand ethos, we train our AI to emphasize brand consistency across all creative assets. The system is designed to create varied outputs that all convey the required brand messaging."

19:27 – Max:
"We also ensure that if our generative AI doesn’t produce the desired results—for example with images—it’s supplemented with stock or client-supplied photos. Our goal is always to achieve on-brand, varied, and high-performing creative."

20:36 – Max:
"Next, let’s discuss consumer research and persona building. Our workflow begins with importing first-party data or survey results to create detailed consumer profiles. Predictive modeling helps us generate around 15,000 insights related to values, motivators, media habits, and purchase behaviors."

21:38 – Max:
"Once we understand our target personas, we use generative AI to produce multiple copy and imagery variations that resonate with each segment. These generated outputs are then preapproved through our internal process before being tested in the market."

22:47 – Max:
"Our next step is pre-launch optimization. Using predictive AI tools, we simulate ad performance before launch, scoring each creative. This process gives us a head start by optimizing headlines, readability, and calls to action based on historical ad data, achieving up to 89% accuracy."

23:45 – Max:
"After launching, our system continuously monitors performance. Computer vision analyzes image elements and text while tagging styles and persuasive techniques. These insights are fed back into the loop to further refine the creative in real time."

24:45 – Max:
"I want to break down the three main AI technologies in our workflow: generative AI (to create copy, images, and video), predictive AI (to score and optimize creative before launch), and computer vision (to analyze and tag ad components for further insights)."

26:53 – Max:
"One of the most exciting aspects is that even if you’re a small agency without proprietary tools, you can curate your tech stack from available AI solutions. Our role has been to research and bundle the best tools into our workflow so that it continuously evolves as new technology becomes available."

29:06 – Max:
"While I can’t reveal the specific companies in our tech stack, its design is proprietary. The stack is dynamic and adapts as AI advances, ensuring that our creative optimization process always remains cutting edge."

30:09 – Max:
"Imagine a future where API-level integration and autonomous agents coordinate across different AI tools. That is where Ventas AI is heading—fully automated creative production that continuously adjusts based on performance data."

31:15 – Max:
"Now, I’ll share an example from pre-launch optimization. A client initiated a design for a report download. The original creative scored 74. After implementing recommendations—such as larger headlines, improved readability, and a bolder call to action—the score improved to 80, resulting in a 27% increase in click-through rate and a significant drop in cost per download."

32:19 – Max:
"Another example: Our computer vision analytics identified that images featuring groups of people had a 9.2% predicted impact on conversions. Based on that insight, we produced more ads featuring group travel, which then contributed to a 66% drop in cost per lead."

33:20 – David (asking):
"Regarding creative specifics—when AI identifies elements like group travel or a subtle design tweak—how universal are these rules across campaigns? For example, if the AI spots that repositioning a download button works, is that applicable for every campaign?"

34:23 – Max:
"The insights serve as guidelines. While certain principles, such as larger headlines and clear calls to action, hold true generally, each campaign is unique. Our system detects nuanced differences through granular testing, enabling adjustments tailored to each client’s audience."

35:26 – Max:
"The AI continuously suggests variations—from facial expressions (like smiling versus frowning) to subtle changes in imagery. Even elements as minor as the color contrast on a download button can have a significant impact. The system reveals these observations, letting us iterate rapidly."

36:28 – Max:
"Each new ad batch incorporates these insights. By testing even small adjustments, we learn more granularly about what impacts performance. The AI’s ability to suggest these tweaks quickly is one of the most powerful aspects of Ventas AI."

37:35 – David (commenting):
"I see this system adapts creative elements rapidly. It even helps you notice things you might otherwise miss, like a barely visible download button. Could you elaborate on how such insights become part of the creative cycle?"

38:41 – Max:
"Absolutely. The system aggregates data on ad performance—copy style, image composition, even the prominence of key elements. It then identifies trends and recommends replicating successful attributes across variants. Over time, these adjustments become part of our best practices for each campaign."

39:38 – Max:
"These tools constantly evolve. For instance, our scoring mechanism, which initially focused on obvious metrics like image area, has grown more sophisticated. Now it also analyzes copy, tone, and other subtle creative factors, ensuring continuous improvement in ad performance."

40:43 – Max:
"Questions often arise regarding the future of AI-assisted creative. There is potential for autonomous creative decisions where machines learn and evolve without constant human intervention. However, human oversight remains essential to ensure alignment with strategic brand messages and to make final adjustments."

41:46 – Max:
"Even in tasks like optimizing a download button or adjusting facial expressions, the AI offers valuable insights. The more ads we test, the better the system becomes at detecting minute improvements. This rapid testing drives efficiency and performance."

42:48 – David:
"How does this approach differ when adapting creative for different platforms—say, Meta versus Pinterest or LinkedIn?"

44:01 – Max:
"Each platform has its own best practices. Our system is trained to generate content that meets the specific requirements for each platform. For instance, ads on Meta have character count and layout constraints that differ from those on display networks. We adjust copy, imagery, and even light motion elements accordingly to maximize performance."

45:08 – Max:
"Sometimes the same creative may be adapted across platforms; other times, a tailored approach is necessary. Our workflow supports both scenarios, ensuring the content is optimized for each platform’s unique environment."

46:10 – Max:
"Here are some examples: This modular design allows us to adapt headlines dynamically and vary images based on the target program. For a campaign in horse and large animal medicine, we produced variants with different images and a distinct yellow tag to highlight specific programs."

47:19 – Max:
"Another example is a campaign for exploring locations like Machu Picchu, where we developed unique variations for each of the 20 different locations the client offers. We also generate creative using generative AI—sometimes producing entirely synthetic images that diffuse a boardroom setting or keynote speaker setup."

48:19 – Max:
"In some cases, we also integrate light motion or even video elements depending on the platform and creative strategy. The system’s flexibility enables both static and dynamic formats as needed."

49:28 – David:
"Are you extending these capabilities to search ads, in addition to paid social and programmatic channels?"

50:32 – Max:
"Yes, we are expanding Ventas AI to cover search ads. Many of our current tools focus on paid social and programmatic, but we’re actively integrating copy generation and optimization for search campaigns as well."

51:35 – David:
"Do you still incorporate human intervention during ad creation if something appears off visually or stylistically?"

50:32 – Max (continued):
"Absolutely. While the AI drives the workflow, human oversight ensures that creative output meets quality standards. We often validate AI recommendations and intervene when necessary, especially because this is performance-driven marketing."

51:35 – David:
"There's a debate about performance marketing versus branding. Some argue that focusing solely on lower-funnel performance might risk brand identity. How do you address that balance?"

52:38 – Max:
"Our focus with Ventas AI is indeed performance marketing—driving specific actions like conversions. However, performance marketing also influences brand perception. We maintain quality and brand consistency through rigorous creative reviews and by aligning all outputs with the brand’s established voice."

53:47 – David:
"Branding always plays a crucial role. Your performance efforts inevitably shape brand identity. It’s a calculated risk, blending immediate performance with longer-term brand equity."

54:50 – Max:
"Our system is designed for auction-based media, where creative performance directly affects exposure. By aligning our creative to the platform algorithms, we ensure that performance enhancements simultaneously bolster brand presence."

55:56 – David:
"Thanks for all the education and inspiration, Max. You've opened up new ways of thinking regarding creative optimization. I appreciate the detailed breakdown and real client examples that illustrate Ventas AI’s impact."

Closing Remarks – David:
"Thank you, Max, Stuart, and the team for contributing to this discussion. We’ll see you in the community and at our upcoming sessions, including a showcase with Textbook Ventures. Thanks everyone for joining today—see you next week!"

Additional Comment – David (in chat):
"One viewer asked, 'When will this video be available?' It will be posted by the end of the week. Thank you all for your engaging questions and comments!"