Rafa Flores, Chief Product and Growth Officer at Treasure AI, led the company's pivot from a traditional customer data platform to an "intelligent" CDP - one that embeds an AI agent foundry and a plain-English orchestration layer. At the same time, he used a custom machine learning model to predict his own epileptic seizures, applying the same iterative product thinking to a high-stakes personal challenge. His dual-track work shows how product development principles can span enterprise systems and individual health.
What makes a CDP intelligent
Flores said the intelligence lies in the AI agent foundry, a tool for building and deploying agents directly inside the CDP. The platform also runs machine learning models for tasks like next-best-action recommendations. Customer feedback drove the creation of Treasure AI Studio, an orchestrator that accepts natural language commands.
"You can literally go into the product and say, 'Hey, I want to run a campaign for cross sales across the following pet care segments, but I'm not sure if this is the best segment.' Or, 'Rebuild the segment for me and tell me why this is the best segment,'" Flores said. Users can still operate the software through a traditional UI if they prefer manual control.
Building this kind of agent-driven orchestration inside a product is a skill set covered in AI Agents & Automation Training.
Product development at speed
Flores returned to Treasure AI in 2025 after years at larger firms. The CEO and board, who are close friends, asked him to help build fast. "My DNA is just to build fast and fix things. I've been in corporate America. I don't do well in corporate America because I get bored," he said. His focus is on how quickly a trusted product-engineering team can deliver value.
For product teams moving from data foundations to AI orchestration, structured approaches can speed up the transition. AI for Product Development Courses & Certifications offer frameworks for that kind of shift.
Automation replaces drudgery, not engineers
Asked whether engineers fear job loss, Flores was clear. "We don't believe AI is going to take jobs away," he said. "We believe that AI is going to do a lot of the stuff that people don't like doing and shouldn't be doing, so they can focus on what makes them the strongest." Engineers still review all automated outputs and write production-quality, safe code - that remains their core value.
Using a personal model to manage epilepsy
In 2022, a serious injury triggered epilepsy. At one point, Flores had up to 20 seizures a day. A neurologist told him to identify his auras and triggers. Whenever he sensed a metallic taste - a pre-seizure signal - he took emergency medication and logged the day's activities, screen time, and noise exposure. He fed those notes into a custom machine learning model, not an LLM, that now predicts risk levels and sends alerts to his phone.
"Without that support system, aka the tech I had for myself, I probably wouldn't even be working today," Flores said. He sees a future non-profit market for such a tool, though his immediate goal is to help others understand epilepsy.
Why this matters for product development professionals
Flores' experience offers two practical lessons. First, embedding AI agents and natural-language orchestration can make a data foundation's value concrete to customers. Second, the same iterative, data-driven mindset that ships enterprise features can tackle deeply personal problems. Product leaders can apply these patterns by prioritizing fast build cycles, automating repetitive work to free engineers for high-impact code, and staying open to uses of AI that sit outside the product roadmap.
Your membership also unlocks:
