Is CDP dead as we know it?

For the last four years, I've lived and breathed Customer Data Platforms. I've coded CDP integrations, evangelized their benefits, and spent countless hours thinking about how to make them more useful. I genuinely believed CDPs were the silver bullet we'd all been waiting for. After years of wrestling with customer data scattered across dozens of systems, the promise of a unified view felt revolutionary.

But here's the thing about working in this industry long enough: you start to see patterns. What once seemed groundbreaking begins to feel... ordinary. And that's exactly what happened with my relationship with CDPs.

The Birth of Tracardi

I was looking for a new project to dive into, something that could make a real difference. After reading countless articles about the state of customer data, one thing became crystal clear: siloed data wasn't just a technical headache—it was a massive business problem. But here's what struck me most: the data itself was only half the story. What we really needed was data that could trigger automation, creating seamless customer personalization and experiences.

That's how Tracardi was born—a platform I built to marry automation with customer data. It wasn't enough to just unify information; we needed it to act intelligently, to respond, to create those personalized moments that actually move the needle. I was completely sold on the vision of CDPs not just as data repositories, but as engines of customer experience.

The Wake-Up Call

Fast forward a few years, and I'm having a different kind of conversation. I started facing a nagging question: why do CDPs limit themselves to just customer data? Why not all data? And standing on the edge of an AI revolution, this limitation began to feel almost absurd. AI breathes data—it thrives on context, patterns, and connections across every aspect of business. Why were we artificially constraining ourselves to customer data only when the real insights come from understanding the complete picture?

The market had moved while we were busy implementing CDPs. The very tools we used daily—our marketing automation platforms, CRM systems, and analytics suites—had quietly absorbed most of the capabilities that made CDPs special in the first place.

Why I Changed My Mind

Three key realizations shifted my perspective:

The functionality had become commoditized. What once required a specialized platform was now table stakes for any decent customer engagement tool. Why add another layer when your existing tools already handled data unification?

Customer data alone wasn't enough anymore. I started noticing that our most successful campaigns weren't just based on customer behavior—they incorporated real-time business context. Inventory levels, supply chain data, competitive pricing, current promotions. The richest customer profile meant nothing if we couldn't act on it with current business reality.

What I See Coming Next

Here's where I think we're headed, and it's honestly more exciting than the CDP era ever was.

We're moving toward what I call Fact Sourcing Data Platforms—systems that don't just collect customer data but understand it within the complete business context. These platforms combine traditional CDP capabilities with event sourcing, real-time analytics, Semantic Web, and AI agents that can act on insights immediately.

It's not about having a perfect customer profile anymore. It's about having all data with context—the bigger picture of customer behavior plus the insights that derive from this data and context. The platform must behave more like a human: observing, memorizing facts, constantly analyzing, identifying entities that participate in these facts, and activating customers autonomously. We need platforms that don't just acquire data but have an inner process of making sense of that data.

The Personal Takeaway

Am I saying CDPs are dead? Not exactly. But I am saying that the narrow, traditional CDP—the one I used to champion—feels increasingly irrelevant.

What we really need now is systems that leverage every possible technology to make company data actionable. We should be using RAGs, Knowledge Graphs, search engines, and AI to get the full picture of a customer. Segmentation shouldn't be based just on the data we've collected, but on all the insights derived from what we know.

Here's the thing: CDPs should know what they know and use this knowledge to help marketers. Most CDP users have no idea what data they've accumulated over time or how to use it effectively for customer segmentation. Imagine having 100 million events collected over years—how do you even begin to know which ones matter?

And here's another reality we can't ignore: data ages incredibly fast. Information that's more than a year old tells us very little about what a customer is interested in right now. Our platforms need to reflect this temporal decay, not treat all data as equally valuable regardless of when it was captured.

That's a much more complex and interesting problem to solve. And honestly, it's the challenge I wish I'd been tackling from the beginning.

What next?

So what's next for Tracardi? We're writing a completely new version from scratch, and this time we're thinking fundamentally differently about data.

Instead of just collecting events, we're moving to observations—rich, contextual records that capture the entities (and their data) involved plus the relationships between them. These relations can be described as events (what happened), insights (what we noticed), facts (factual data), or any other meaningful connections. We're organizing all of this into a knowledge graph with semantic meaning of what happened and who participated. It can be as simple as "Product X was delivered to Person Y" or "Product price has changed." Now it's not only about the customer, is is about all entities involved and action. Picture this. If we know that "product X price have changed" and "Person Y is interested in Product X or alike" and "Customer X is price sensitive" that's a huge insight that the system could use.

The goal is to make data more human-like. Instead of drowning in millions of disconnected data points, we want to create a system that understands context and relationships the way humans do. We're using AI not just to process this data, but to actively derive insights from it.

And perhaps most importantly, we're concentrating on simplicity. The complexity should be hidden under the hood, while the interface and insights remain elegantly simple and actionable.

This is what I should have been building four years ago. But sometimes you need to go through the journey to understand where you really need to end up.

Starting Small, Thinking Big

We're taking a different approach this time. Instead of diving straight into enterprise complexity, we're starting with something simple: a platform made for personal use.

We're launching a new product called AiRembr that will serve as our test bench for these ideas. Here's what we noticed: collecting data for a single user is remarkably similar to collecting customer data for a company. If we can excel at the personal level—helping individuals make sense of their own digital lives—then we can take that refined experience and scale it to the company level, where we'll use it to memorize customer behaviors and build intelligent segments.

AiRembr will be completely free for personal use and will exist as a separate product from Tracardi. Think of it as our laboratory for proving that this human-like approach to data actually works.

Then, after we've refined the concept through AiRembr, we'll launch Tracardi 2.0. This new version will use the same engine but scale it to handle thousands of customer memories—memories that companies can use to fuel their marketing and customer personalization efforts.

It's a more humble beginning than my original CDP evangelism, but I think it's exactly the right way to build something that truly works.

If you're interested in our journey and products, please sign up for this blog so you don't miss new ideas and the steps we're taking along the way.