Insight
Designing AI Products
The user interacts with AI. When the model becomes part of the product — not the tool you built it with — you stop designing screens and start designing the collaboration between a person and a system that works out loud.
The old flow: deterministic screens
Click a button, upload the PDF, wait, view the results. A straight pipeline where the software moves data and the person just waits — every step the same input, same path, same output.
The new flow: collaboration
The same task reshaped: the AI starts extracting the moment the file lands, streams its progress, flags the fields it's unsure about, and hands them to the person to fix and approve. Same PDF, a completely different thing to design.
The real design surface: the questions
When should the AI speak, and when stay quiet? How much confidence should it show, and how do you make uncertainty visible? When does a human approve? These interaction decisions — not the screens — are where the design work moved.
A few years ago, when a producer needed to pull the numbers off a loss run, the work was a straight line. Click the button, pick the file, watch a spinner, come back to a table. The software's job was to move data from the PDF into the system; the producer's job was to wait, then check every row by hand. Each screen was deterministic — same input, same path, same output. You designed the screens and the flow between them, and that was the whole job.
Now the AI sits inside the product. Drop the same loss run and it starts reading the second it lands — carrier, effective dates, loss amounts — streaming the fields in as it finds them. It marks the three it's not sure about instead of pretending it's certain. The producer fixes those, approves, exports. Same task, but there's nothing linear left to lay out. You're choreographing a back-and-forth between a person and a system that's working right there next to them.
It's the same shape outside insurance. A journalist researching a piece drops in her notes and sources; the AI pulls the relevant quotes, drafts a passage, and flags the claims it couldn't stand behind. She rewrites the lede, cuts what's wrong, keeps what earned its place. Different work, identical collaboration — the person and the machine trading moves, the system showing its uncertainty, the human holding the final call.
And that back-and-forth is made of decisions that didn't exist a few years ago. When should the AI speak up, and when should it stay out of the way? How much confidence should it show — and how do you make uncertainty visible without making it scary? What runs automatically, and what waits for a human to say yes? How does someone climb back out when the model gets it wrong? None of these are screen-layout questions. They're interaction-design questions, and they're new.
The tempting move is to treat the AI as a feature you bolt onto an existing screen — a button that summarizes, a panel that suggests. But the moment the model is doing real work in front of someone, the thing you're designing isn't the screen anymore. It's trust: whether the person believes what they're seeing, knows when to step in, and still feels in control of something that's partly out of their hands. Get that wrong and the smartest model in the world feels like something you can't rely on.
Design the collaboration, not the screen.