AI Ventures · Applied, not theoretical
Three ventures, each started from an operational problem I could state in one sentence. Below: the problem, the product, the evidence so far, and the next responsible step.
Six conditions decide whether an AI initiative survives contact with an operating business — and every one of them is enterprise-systems work, not model work.
Experience proves
Enterprise architecture and delivery judgment — security models, integration estates and release governance at Fortune 500 scale.
AI Ventures proves
Practical experimentation — building, evaluating and stress-testing AI workflows against problems real people actually have.
Together
An advisor who can look at the systems you run today and tell you honestly where AI belongs next — and where it does not.
Problem · product · evidence · next step
digital product passport
Problem
Paper manuals get lost; the helpdesk answers the same questions all day.
Product
Scan the QR, text a question, get an answer drawn from that product’s own manual — low confidence escalates to a human.
Evidence
End-to-end path working in 11 languages against a fixed question set. No deflection rate published yet.
Next step
One manufacturer, one product family, call volume measured before and after.
field capture to clean record
Problem
Field work arrives as photos and voicemail, then someone re-types it a day later.
Product
Capture naturally on site; the model structures and routes it; a named human confirms before it commits.
Evidence
In build. No adoption or accuracy data yet — stated plainly rather than dressed up.
Next step
One operator, one workflow: re-keying time removed against errors introduced.
a place report you can check
Problem
Place decisions run on marketing and opaque scores that cannot be audited.
Product
Public data scored per dimension and re-weighted around the reader — every claim carries its source and a confidence band.
Evidence
A full Austin-metro report is generated and viewable, 17 cities scored.
Next step
More metros, with source-quality flags per region.
Working artefacts from the three ventures — what was actually built, what it runs on, and how far along it is. Each entry links to the thing itself where there is something to open.
Advisory
