I’ve spent more than two decades on the data side of organizations first as a consultant, later running technology and digital transformation inside government, and eventually building analytics practices for companies across industries. For most of that time, the rule was simple: whoever had more data had the advantage. Collect more, process it faster, extract more insight than the competition.
That rule is starting to break.
AI has made intelligence abundant. Any organization can now generate content, detect patterns, predict behavior, and infer things about people they never explicitly said. What used to take a skilled analyst weeks now takes minutes.
But abundance of intelligence doesn’t solve the real problem. It exposes it.
The real problem is trust.
More data isn’t automatically better data.
I’ve seen organizations sit on enormous datasets (transactions, CRM records, behavioral logs, inferred preferences) and still make bad decisions, because nobody could tell you where half of that data actually came from, how old it was, or under what conditions it was collected.
AI doesn’t fix that. It amplifies it. A powerful model built on unreliable inputs doesn’t produce better decisions; it produces wrong conclusions with more confidence and more speed than before.
So, the advantage in this next phase isn’t going to belong to whoever has the most data. It’s going to belong to whoever can trust theirs.
Provenance is a quality problem, not just a legal one.
For most of my career, organizations kept these questions separate. Accuracy sat with the data team. Privacy sat with legal. Consent sat with compliance. Three different conversations, three different owners.
That separation doesn’t hold up anymore.
If you don’t know where a piece of information came from, when it was given, or under what terms it can be used, I don’t think you can honestly call that high-quality data, no matter how clean it looks in a dashboard. Provenance isn’t a compliance checkbox. It’s part of what makes data usable at all.
Inference is powerful. It’s still a guess.
AI can predict what someone is likely to buy, watch, or do next without ever asking them. That’s genuinely useful. But it’s still an interpretation of behavior, not a fact about a person.
I think about this every time I see a company trying to explain why a customer left. They’ll run the churn model, look at the usage drop-off, check the support tickets, and land on a plausible story. Price sensitivity. A competitor. Changing needs. All reasonable guesses.
Or they could have just asked.
Declared information does something inference can’t: it tells you the why, not just the what. And it gives you a way to check whether what your models are inferring about people is right.
People are still on the outside of this.
This is the part that bothers me most, honestly. Organizations spend enormous effort reconstructing who someone is from the trail they leave behind while rarely giving that same person any real role in the exchange. All the analysis happens about them, not with them.
I don’t think that’s sustainable, and I don’t think it’s necessary. The next stage of this isn’t less AI or less inference. It’s building a real channel back to the person the data is about, one where they understand what’s being asked, why it matters, and what they get out of participating.
That’s not a technical problem. It’s a design and trust problem.
Why do I think this matters commercially, not just ethically.
Trust gets talked about as a nice-to-have, something companies should do because it’s the right thing. I’ve come to think it’s an economic lever.
When the data you’re working with is recent, well-contextualized, and given intentionally, you stop drowning in noise. You ask sharper questions. You combine what you observe with what people tell you. And people are more willing to share when they understand what’s happening and why.
That changes the math. The goal stops being “collect more” and becomes “understand better.”
Where this leaves me.
I’ve spent most of my career helping organizations get better at extracting value from data, in private sector, and later running technology for government at a national level, where the stakes of getting this wrong were a lot higher than a bad marketing segment. What I kept running into, in every one of those roles, was the same gap: we were good at technical problems and bad at human ones.
That gap is why I started deitta. Not because the world needs another data platform, but because I think the architecture underneath the data economy needs to change, toward a model where the people generating the data are actual participants, not raw material.
AI is only going to get more powerful. Data is only going to keep multiplying.
Organizations that win the next round probably won’t be the ones that collected the most. They’ll be the ones that can answer a harder question: can you actually trust the data your intelligence is built on?
