Work
I have noticed the useful projects usually start with some version of: this should work, but it does not yet.
Sometimes the missing piece is product clarity. Sometimes it is a workflow. Sometimes it is distribution.
Project 1
What was broken
AI could sound confident before the answer was grounded.
What I built
A workflow graph that retrieves evidence, routes it, evaluates it, repairs weak work, and pauses before spending or publishing.
What it shows
The first version looked cleaner than it was. So I fixed the system instead of pretending the output was fine.
Project 2
What was tested
A company assessment through the GTM AI Council product.
What it proves
Positioning, messaging, product marketing, conversion systems, distribution, assumptions, risks, pilots, metrics, and ROI logic.
Limit
The assessment separates known, assumed, and unknown company data. It does not fabricate missing CRM, revenue, or conversion values.
Project 3
What was broken
Education was not enough. Timing, fit, proof, and trust carried more weight.
What I shaped
Brand materials, sponsor materials, investor education, presentation assets, entity materials, financial-model work, and narrative structure.
What it shows
A sponsor, advisor, investor, executive, and operator do not evaluate the same product the same way.
Visit dstprogrampartners.com · Read the DST Program Partners case study
One heroic employee and one heroic spreadsheet is not an operating system.