Switchboards Library · Start with the person, not the tool.

Library Human-centered systems

What is user-centered design?

User-centered design means building around the person who has to use the thing: what they are trying to accomplish, what is getting in the way, and what would create real relief.

6 minute readAI systemsUpdated August 2026

User-centered design is an approach that uses real people’s goals, context, behavior, and feedback to guide what gets built and how it works.

Start with the stuck moment.

Most people do not wake up wanting an automation, an agent, or a dashboard. They want to stop missing important messages. They want to know what needs attention. They want content to stop consuming the whole week. They want the business to feel less scattered.

User-centered work begins there. Instead of asking, “What can this AI tool do?” ask, “Where does this person lose time, energy, confidence, or momentum?” The answer gives the system a job.

Relief before features

A founder says, “I never know what to work on first.” The user-centered response is not automatically a project-management app. It may be a short morning brief that understands her current goals, gathers what changed, and suggests one useful next move.

Four questions keep an AI system centered on the user.

1. What is the person trying to make easier?

Name the burden in ordinary language. “Keep me from dropping balls” is often more useful than “centralize operational data.”

2. What context does the system need?

Useful help depends on the person’s business, preferences, responsibilities, examples, boundaries, and way of working. Without that context, even a powerful model gives generic advice.

3. Where should the person stay in control?

Decide what can happen automatically, what needs review, and what should never happen without permission. User-centered does not mean removing the human; it means involving them at the moments where judgment matters.

4. Does the result change the day?

A system succeeds when the person can make a decision, complete a task, notice a risk, or reclaim attention. Output that creates more sorting is not relief.

User-centered AI does not require prompt mastery.

People should be able to explain what is happening in their own words, ask ordinary questions, and receive guidance that fits the situation. The system can structure the mess, ask for missing information, and help carry the work forward.

That is the practical standard: the technology adapts to the person and the work. The person should not have to become a machine to use it.

Start with the relief

Bring the real problem.
Build around the real person.

Join the waitlist for the next live Switchboards workshop.

Get the first invitation when this opens. Unsubscribe anytime.