How to Choose a Partner to Integrate AI into Your Design Process Responsibly

Choose the partner who talks about your customers and your risk before they talk about models. Then ask them five questions. How they answer will tell you more than their portfolio.

Here are the questions, with what a good answer and a weak answer sound like.

1. What will you do before you build anything?

Good answer: "We'll study the task first, find where people spend time on low-value work, and decide where AI helps and where it shouldn't go."

Weak answer: A demo of a tool they already like.

2. What happens when the AI is wrong?

Good answer: They describe the design of the failure: how the person spots a bad answer, checks it, undoes it, or reaches a human.

Weak answer: "The model is very accurate." Every model is wrong some of the time. A partner who hasn't designed for that hasn't done this before.

3. How will you test for bias?

Good answer: They name the groups most likely to be let down, explain how they'll recruit them, and show you the test cases.

Weak answer: "The vendor handles that." Maybe. But your customers will blame you, not the vendor.

4. What will people be told?

Good answer: A short, plain-language explanation of what the AI does, tested with real users to make sure they understand it.

Weak answer: A legal notice nobody reads.

5. How will it fit our systems and our obligations?

Good answer: They talk through where the data goes, your privacy obligations, what gets recorded for audit, and the effect on performance and page speed.

Weak answer: "It plugs straight in."

Red flags

• They lead with the technology, not the problem.
• They can't show you a project where they decided not to use AI.
• Accessibility and testing are a phase at the end, not part of the start.
• Nobody senior will be on the work.

What you should have at the end

A feature people understand and trust. A record of what it does and who approved it. And evidence, from real users, that it saves time without lowering the quality of the work.

For how we apply this ourselves, read our principles for responsible AI in design.

Related: Innovation coaching and Our approach.