The Pattern
Most companies treat real-time data as a display problem. The data streams through like water through a sieve; the sieve gets credit for being near the water.
Most dashboards tell you what happened. Past tense. A record of things you already missed. Few tell you why it happened. Almost none tell you what to do next. And none of them do it for you.
The harder problem—the one that keeps surfacing across every domain I work in: how do you turn volatile data into durable intelligence? Not better charts. Not more analytics. A system that understands what's happening well enough to act on it.
I'm currently working on problems in this space under a confidentiality agreement. I can't share the client, the domain, or the specifics. But the pattern is worth describing—because almost nobody builds for it.
The Missing Layer
Think about Pandora. Their intellectual property isn't the music—it's the Music Genome Project: a system of definitions encoded as data. Those definitions feed algorithms that power recommendation and advertising. The music is the surface. The genome is the system.
This pattern appears everywhere, the way load-bearing walls appear in every building but nobody photographs them for the real estate listing. Raw data streams that could be transformed into structured intelligence. That intelligence could power multiple experience layers—each feeding on the same underlying reality, each reinforcing the others.
One source of truth. Multiple expressions. The companies that build this own something defensible. The companies that don't are competing on surface.
Why Nobody Builds It
It requires solving three problems at once, and they belong to different departments.
The modeling problem: What states matter? Most real-time data is noise wearing a suit. An intelligence layer has to distinguish signal from noise—and that requires someone who understands both the data and the domain.
The expression problem: How do you connect meaningful states to experiences people care about? A threshold crossing isn't inherently interesting. It becomes interesting when it activates something worth paying attention to. Designing those connections is systems design, not feature design.
The coherence problem: How do multiple experiences feed on the same reality without contradicting each other? One source of truth, multiple expressions—but the expressions have to cohere. This is an architecture problem disguised as a content problem.
Most companies solve one of these. Few attempt all three. The ones that do tend to become the platform everyone else builds on.
What I Can Say
I can't share what I'm working on now. But the rest of this portfolio demonstrates how I think about exactly this class of problem.
Southern Company Gas is a content architecture that produces five websites from one system. Drip is a data system that tells you where to intervene. Grammarly is a system that turns email into a knowledge graph. The derivative is the same even when the function changes. Data becomes structure. Structure becomes intelligence. Intelligence becomes experience.
When I can talk about the current work, I will. Probably far too much.
The Difference
The difference between a thermometer and a thermostat.
If you're sitting on real-time data and treating it as a display problem, you're leaving the system unbuilt. The data streams through. Nothing compounds. The opportunity just sits there, patient as wood.
The intelligence layer is the hardest thing to design because it doesn't belong to any existing role. It's not engineering. It's not product. It's not data science. It's the system underneath all of them.
That's what I design. If you have the data and don't have the system, let's talk.