How I use Claude in my day-to-day design workflow

A year ago, I used AI mostly for curiosity. Today, Claude is part of my daily workflow as a designer and frontend developer. It doesn't make design decisions for me, but it takes care of a lot of time-consuming work around those decisions, giving me more time for the parts of the job that need human judgement.

Turning messy notes into structure

After stakeholder meetings or user interviews, I often have pages of scattered notes. I ask Claude to organize them into themes, list open questions and highlight contradictions. I always review the output against the original notes, but it gives me a strong starting point in minutes instead of hours.

Exploring copy and microcopy

When I'm stuck on a headline, an empty state or an error message, I ask for several variations in different tones. Most won't be right, but reviewing alternatives helps me find the clearest wording faster. I treat AI suggestions as a brainstorming partner, not a final answer.

• Button labels and error messages.
• Onboarding copy in different tones.
• Shorter versions of long explanations.
• Alternative headings for landing pages.

Prototyping with code

For interactive prototypes, I use Claude Code to turn designs into working Next.js components. I describe the component, point it to my existing tokens and components, and review every change. This lets me test real interactions with users much earlier than before.

“AI handles the first draft. I'm responsible for the final decision.”

Reviewing my own work

I sometimes ask Claude to review a flow description or design rationale for gaps: missing edge cases, unclear steps or accessibility concerns. It's like having a second pair of eyes available at any moment.

Where I don't use AI

I don't use AI to replace talking to real users, to make final design decisions or to generate visual designs I can't explain. Taste, empathy and accountability remain human responsibilities.

Prompts I use regularly

Over time, I've developed a small set of prompts for recurring tasks. For meeting notes: "Summarize these notes into decisions, open questions and action items." For copy: "Give me ten variations of this message, from formal to friendly." For reviews: "List edge cases and accessibility concerns for this flow description."

Keeping these prompts in a document saves time and makes results more consistent. I refine them whenever I notice a better phrasing produces better output.

• Meeting summaries with decisions and actions.
• Copy variations in different tones.
• Edge case and accessibility reviews.
• First drafts of documentation.

Protecting confidentiality

Before using any AI tool with client or company information, I check what's allowed and remove sensitive details. Names, financial data and unreleased product information shouldn't be shared without permission. Responsible use protects clients and keeps trust intact.

“Use AI to save time, never at the cost of someone else's trust.”

Putting it into practice

Choose one repetitive task in your week — summarizing meetings, drafting copy variations or reviewing flows for edge cases — and try an AI assistant for it. Write a clear prompt with context, review the output carefully and note how much time it saved.

Refine the prompt over a few weeks and save the best version in a shared document. Share what worked with your team, along with any guidelines for protecting sensitive information.

Key takeaways

• Use AI to remove busywork, not judgement.
• Give clear context in prompts.
• Review every output carefully.
• Protect confidential information.

Used thoughtfully, AI makes designers faster without making them less thoughtful. The key is staying in control of the decisions that matter.