1 Is this you?2 How I work3 The record4 Writing5 Contact
Kevin Holland
Role Fractional chief data and product officer
Now Fractional Head of Product, S2N Health
Before Director of Product, NeuroBlu at Holmusk, through October 2026
Base Charleston, South Carolina
Make messy healthcare data usable for products and AI.
Your AI data product has to ship, and the seat that should own it is empty. I step in as a fractional chief data and product officer: a senior builder and player-coach across data, AI and product. The pilot ships, and your team runs what comes next.
Facts The offer
Question Which urgent data or AI project has no owner, or won’t scale?
Evidence At Holmusk, January through June 2026, brand analysis that took four to eight weeks of custom work per brand became one configured pipeline.
Decision I lead it as your fractional chief data and product officer, and build it with your team.
Limits That case is not end-to-end, and no staffing saving or revenue was measured.
Owner Your team runs it, and the next one, after I leave.
Is this you? Two urgent projects teams bring me. In each, the work can’t wait and the usual options fall short.
An AI data product has to ship, and no one owns it.
Project A pilot-ready AI data product and a roadmap the leadership team approves.
Why now The pilot has dates. The product leader seat is empty, so decisions wait on the acting CEO.
Options Hire a full-time chief product officer. Keep routing product calls through the CEO.
Falls short The search is close to square one. The CEO can’t also run the product.
Email me the pilot that’s stuck
Your team can’t serve every client.
Project Client analysis your team rebuilds by hand, moved into one product the team runs.
Why now Interpretation is so manual the team can’t scale to every client. The current tool is stuck, handed from one person to the next.
Options The vendor tool. An AI assistant on the warehouse. Another data engineer.
Falls short The vendor tool checks the box, but it isn’t what you want. Privacy rules block the assistant on patient data. A new hire still lacks the time to build.
Email me the analysis that won’t scale
How I work Four steps, in this order. I build it with your team.
Clarify Name the decision and the owner.Ground Test the claim against real evidence.Build Ship the smallest system that can change the call.Operate Measure the outcome and close the loop.
The record Led product work across NeuroBlu’s $15M ARR platform business, including $3M in standalone subscriptions.
CensusChat architecture: a plain-English question enters an MCP validation layer (table and column allowlists, row caps, dangerous-pattern blocking) before reaching DuckDB. Output is a sortable table and chart, under two seconds. Anything that fails the gate never reaches DuckDB.
CensusChat
A text-to-SQL demo is easy. A text-to-SQL product a healthcare
org will point at data is a security problem wearing a chat interface.
INPUT
A question in
plain English
“How many adults over 65
in Cook County?”
THE KEY DECISION — MCP
VALIDATION LAYER
Generated SQL is
never trusted
Table and column allowlists
Row caps
Dangerous-pattern blocking
ANYTHING THAT FAILS THE GATE
never reaches DuckDB
OUTPUT
A sortable table
and a chart,
under two seconds
3,144 counties ·
239,741 block
groups · 84 variables
CensusChat architecture: a plain-English question enters an MCP validation layer (table and column allowlists, row caps, dangerous-pattern blocking) before reaching DuckDB. Output is a sortable table and chart, under two seconds. Anything that fails the gate never reaches DuckDB.
CensusChat
A text-to-SQL demo is easy. A
text-to-SQL product a healthcare org
will point at data is a security
problem wearing a chat interface.
INPUT
A question in plain English
“How many adults over
65 in Cook County?”
THE KEY DECISION —
MCP VALIDATION LAYER
Generated SQL is
never trusted
Table and column allowlists
Row caps
Dangerous-pattern blocking
ANYTHING THAT FAILS THE GATE
never reaches DuckDB
OUTPUT
A sortable table and a
chart, under two seconds
3,144 counties · 239,741
block groups · 84 variables
CensusChat. A diagram of the system, not a product screenshot.
Brand analysis rebuilt for every request became a configured pipeline.
About 45 seconds per configured drug from the initial pipeline. That is initial-pipeline output, not an end-to-end rebuild time.
Read the case study
Bring the analysis that won’t scale
If your team can’t serve every client because the analysis is rebuilt by hand, or an AI pilot has no owner, the next step is a direct conversation.
Newsletter
The Data Product Builder
What I’m building, what broke, and what I’d do differently. Healthcare data
products and decision infrastructure.
No spam. Unsubscribe anytime.
