We’re entering an interesting phase of AI adoption: the bottleneck is increasingly not access to models, but access to reliable context.
A powerful model can summarize, predict, and generate but if the underlying data is incomplete, outdated, duplicated, or poorly governed, the intelligence of the system is limited by what it can actually know.
That’s something I’ve become particularly conscious of through my work across technology, sustainability, verification, and data.
In sectors where decisions depend on supplier information, emissions data, compliance records, or other operational inputs, “AI-powered” means very little if the underlying information cannot be trusted.
I think the next competitive advantage in AI will increasingly come from data infrastructure and context, not just better models.
The smartest system is only as good as the information it can reliably access.
