Your data is the bottleneck, not the model
As frontier models commoditise, the firms that pull ahead are the ones whose operational data is clean, connected and findable.
Source: Tech Monitor · Summary and commentary by Sticklebrick
Coverage this month underlines a shift practitioners have felt for a while: the model is rarely the constraint. Messy, siloed operational data is what stops an AI project delivering.
For an established business this is oddly good news. The advantage moves away from who has the biggest AI budget and towards who knows their own operation — exactly the ground incumbents stand on.
What to watch for is scope creep dressed up as a 'data platform'. You do not need to boil the ocean; you need the handful of records an AI use case will actually read, in a state it can use.
Sticklebrick's Data Readiness Score exists for precisely this — a fast, plain-English measure of how ready your data is for a specific use, so you fix what matters and skip what doesn't.