Insights
Most problems I write about have the same shape: the tools exist, the talent is there, and something structural is still getting in the way. These posts are where I work through what that something is. In org design, in AI deployment, in the gap between a dashboard and a decision.
If something here connects with work you are doing or a problem you are working through, my contact is below.
An agent on the lake is fine
Pointing an agent at the lakehouse works for a lot of what agents do. It is not the ceiling. The bridge worth thinking about extends the lakehouse's semantic layer to the integration plane, so an agent operates on the same governed entities whether the data is at rest or in flight.
The Operating Model AI Is Building Toward
Three roles, three contracts, one coherent system: someone who owns the data layer end to end, someone who deploys capability directly into the business, and someone who governs what gets built and where. Most enterprises have none of them clearly defined.
The Product Manager Nobody Has Hired Yet
Every enterprise spends millions on platforms with capabilities nobody has turned on, while engineers build custom solutions for problems the company already owns. The missing role is not more engineers.
The Engineer in the Room
Central AI teams produce proof-of-concepts that never deploy. Business teams submit requests that sit in backlogs for quarters. The problem is not capability. It is distance. Here is the structural fix.
Two Roles, One Job, Nobody Owns It
Most data teams split pipeline work and modeling work across two roles with a handoff in the middle. That handoff costs weeks, loses context, and produces datasets nobody fully owns. There is a better structure.
Data Agents and the Multi-Model Future
The data platform is the right place to build intelligence. Not because any one vendor's models are best, but because the governed data is already there. Build the tools, validate the foundation, earn the trust. Then let the agents run.
Your Data System Should Have an Opinion
The next generation of analytics doesn't wait to be asked. It monitors, detects, recommends, and acts. But agentic AI without a governed foundation is automated chaos.
Stop Bolting AI Onto Bad Data
The companies getting value from AI aren't the ones with the best models. They're the ones who built the foundation first and introduced AI at the decision layer, not the data layer.
What an Agent Is (and Isn't)
A chatbot gives you a bad answer and you shrug. An agent gives you a bad answer and someone acts on it. The industry is calling everything an agent, and the gap between the demo and reality is where executive decisions go wrong.
Nobody Agrees on What a Customer Is
Before you build a dashboard, a forecast, or an AI agent, you need to answer a question most companies skip: what are the nouns? Entity design is the unglamorous foundation that everything else breaks without.
Your Dashboards Are Wrapping Paper
A dashboard that doesn't change behavior is decoration. The difference between a metric and a decision tool is whether anyone does something different on Monday because of what it showed them on Friday.
Why Most GTM Forecasts Are Fiction
If your forecast doesn't change every day, it's not a forecast. It's a report with aspirational formatting.
From Data Strategy to Decision Systems: What It Takes
You can have a modern data stack, clean pipelines, and a warehouse that passes every audit, and still have an organization that cannot agree on what happened last quarter. The tooling is not the problem. The decision system is the problem.
Analytics Pillars Create Information Asymmetry by Design
Organizing analytics teams around business pillars doesn't create alignment. It creates silos with authority over people who can see more than they can.
The Case for an AI Portfolio Strategy
The organizations that win in AI aren't the ones who picked the best model. They're the ones who built the architecture to deploy and govern multiple models as the market moves under them.
Saying No, Specifically and Publicly
A vague no costs you twice. You spend the political capital of pushing back and get none of the protection a real no provides. Here's what a real no sounds like.
Data Is Just Data. The Hard Part Starts After the Pipeline Works.
The tooling is commoditized. A competent team can stand up a modern stack in a quarter. The hard part is what happens the day after the pipeline works. And it has nothing to do with the data.
What I Build When the Budget Is Frozen and the Board Still Wants Growth
When headcount is off the table and the business still expects results, the data leader's job is to connect every dollar of analytical work to a number the board already watches.