Matthew Niederberger - Martech Therapy (Page 8)

Letting go of control

What agentic AI can teach Martech about trust, autonomy, and actual intelligence

I used to think I understood AI agents. In Martech, we talk about them all the time. Tools that take action based on data, automate tasks, score leads, and optimize send times. That’s what we call an agent, right? A helpful script or model that executes a narrow task faster or more accurately than humans can.

Then I had a conversation with Aampe’s CEO Paul Meinshausen on my podcast Couch Confidentials, and something clicked. Paul introduced a distinction I hadn’t fully grasped. An AI agent follows instructions. Agentic AI operates with goals, persistence, and the ability to

marketing technology

Contextual CDPs: Composable, not chaotic

Part 5 of 6: Contextual CDPs need lightweight structure, not heavyweight rebuilds

By now, we’ve explored what context can unlock in a CDP, from dynamic profiling to real-time orchestration and trust. We’ve talked about layering memory and inference without blowing up your current stack. So in this fifth part, let’s look at what happens when you want to make these contextual layers stick… without making everything feel like a pile of duct-taped services.

Because that’s the risk. Composability gives us incredible flexibility, but without some architectural discipline, it becomes hard to explain, harder to maintain, and impossible to govern.

Let’s explore how to structure context-aware systems in

aampe

Agentic AI & the art of letting go

When control gets in the way of progress

Paul suggested the idea of doing something different. No intro. No big kickoff. Just two people talking.

That’s how this episode with Paul Meinshausen (CEO at Aampe) started and I’m glad it did. Because I fumbled at the beginning (you can skip ahead in the podcast...).

Feeling like a one-trick pony, I tried to steer the conversation toward my usual topics: CDPs, AI, use cases, and organizational challenges. But it quickly became clear I was trying to force an idea, a vision, onto something I hadn’t fully wrapped my head around yet. Paul helped me take a

cloud data warehouse

If the warehouse can think, what’s left for the CDP?

A critical reflection on Databricks, data gravity, and what comes next.

Nope, this isn’t an analysis of a press release. We’ve seen all Databricks partners do that generously all over LinkedIn. It’s a reflection, a personal one (I had some spare time over the weekend), on something that’s been building quietly in the background and what it means now that someone has taken a concrete step forward. That someone is Databricks.

Last week, Databricks launched something called Data Intelligence for Marketing. And sure, you could skim the headline and file it under “AI feature drop.” But that’s not what caught my attention.

What stood out was

tealium

Are MCPs the USB-C of AI? Tealium Thinks So

A conversation with Tealium’s Sav Khetan and BJ Allmon about MCPs, AI integration, and the role of CDPs in a multi-agent world.

In this episode of Diagnosis Demo, I sit down with Sav Khetan and BJ Allmon from Tealium to explore their recent implementation of MCPs (Model Context Protocols). We discuss what MCPs mean for real-time AI integration, how Tealium positions itself as an AI enabler, and why the future might just belong to multi-agent architectures. We also demo a real use case that shows how fast and powerful Tealium's setup really is.

Watch to see:

A working demo using Tealium’s Moments API and MCP integration in a simple chatbot
The difference between MCP and A2A protocols
Why data orchestration, not

marketing technology

Contextual CDPs: Building the bridge

Part 4 of 6: What connects today's CDPs to tomorrow’s agent-driven context

If the first three parts of this series explored what’s possible when context becomes the driving force behind customer data, then this one asks a more grounded question:

How do we actually get there?

I’ve spoken with teams who are genuinely excited by the idea of using memory, inference, and context to make their CDPs smarter. But many feel stuck between inspiration and implementation. They see the potential, but don’t quite know where to start.

Let’s be honest, the martech stack is already complex enough. The last thing anyone wants is to bolt on a new

Posts by Matthew Niederberger