AI Metadata Enrichment Pipeline
A governed AI enrichment pipeline processing millions of assets, with every output traceable to the prompt and model configuration that produced it.
Read itHello from Minneapolis
I build systems
people trust.
Hi, I'm David LaHaye. I lead enterprise content and AI platforms at Target, after a career that ran through design, software, product, and platform leadership. The system is the puzzle I love. The people who depend on it are the point.
Principal Product & UX Leader
Minneapolis, Minnesota
How much time do you have?
Now
Leads enterprise content and AI platforms at Target
Built across
Design, software, product, platforms, and now AI
One rule
Understanding the problem is the work
Hello again
If we've worked together, you already know the pitch. Skip it. It's good to see you.
The short version of what's changed: I'm at Target now, leading enterprise content and AI platforms. I'm writing more than I used to, mostly about why large systems are hard to trust and what to do about it. Still in Minneapolis, still golfing, still learning.
The work, in plain terms3 min
Fragmentation
is the tax.
Every large organization pays it. Three teams build the same approval flow. A field means one thing in one tool and something else in the next. Search returns last year's version. Nobody chose any of this. Each decision was reasonable on its own, and the sum of them is a daily cost that rarely shows up in a budget. I find where it is being paid and remove it.
What a team says
What is usually underneath
“Our search is broken.”
A metadata problem.
“Adoption is low.”
A product problem.
“The AI pilot never shipped.”
A governance problem.
“Nobody knows who owns this.”
An operating-model problem.
“We have four tools that do the same thing.”
Four reasonable decisions, made separately.
Understanding the problem is the work. The fix is usually smaller than the diagnosis, and it holds because the people inside the system chose it.
How I approach itA short note, since you're staying a while
One of the more curious things I've noticed is that nobody sets out to build a confusing system. A tool gets bought because a team was stuck. A field gets added because a report needed it. A review step shows up because something went wrong last spring. Each choice is reasonable on its own, and the sum of them is a place where nobody can find the current version of anything.
That's not a criticism of anyone. It's what reasonable people do when they solve the problem in front of them. The work I like is standing far enough back to see the whole thing, then making it simpler for the people inside it without telling any of them they were wrong.
I live in Minneapolis. I'm a dad, a golfer, and still, happily, a student of all this.
If you read one thing3 min
Adoption Is a Product Problem, Not a Change Management Problem
People do not resist new systems because they are unfamiliar. They resist because the new system has not made their work easier yet. That is a product gap, not a communication gap.
Read itAI Governance Is an Organizational Design Problem
AI governance failures are rarely technology failures. They happen in organizations that were never designed to govern intelligent systems.
Read itThat's the thirty-second version.
That's the three-minute version.
Ask, without the small talk
Have a question?
Ask it.
Ask is a research surface built on the essays and case studies here. One question, one answer, with sources. It declines anything off topic, and it does not pretend to be me.
Open AskOr press / anywhere
If you are here for a reason
If you are navigating complexity, scaling platforms, or aligning teams around shared systems, we are likely speaking the same language.
Read the resume. Additional context on request.