Product design leadership for AI people can trust

Until recently, trust in enterprise software was someone else’s job. Sales vouched for the product. Customer Success absorbed the failures. Data and Security teams handled the technical integrity. The buyer trusted the people; the end user tolerated the software.

That model breaks when the product itself starts making judgment calls.

Agentic workflows and chat interfaces have changed the relationship. A frontline manager acting on an AI recommendation isn’t trusting a brand or an account team. They’re trusting the product, in real time, on a decision that affects their work, their team, their numbers. There’s no human between them and the model. Trust has become a UX problem, and an urgent one.

In practice, this means three moves. Understand user intent more precisely than before, not just what they asked for, but what they’d accept if the product got it wrong. Pattern transparency and feedback into the interface itself, so users can read the product’s reasoning without having to ask for it. And surface ROI in-product, so trust compounds into adoption rather than stalling at tolerance.

Intent → Transparency → Trust → Adoption → Growth. This is the loop. It’s how software earns its place in workflows that can’t afford to be wrong.