AI Design Tools in 2025: What Actually Worked

Vaibhav Sisinty tested ten AI design tools throughout 2025 and found that while everyone obsessed over AI coding tools, design tools quietly got good. His newsletter, Staying Ahead with AI, breaks down which tools won in specific categories and which ones failed to deliver.

The year saw major releases. Figma launched Make in May. UX Pilot arrived with actual user understanding. Google released Stitch as an experiment. Most designers stuck with Make, but those who explored further found advantages.

Full product design

Stitch generates entire products or complete user flows from a single prompt. Ask for a dashboard and it returns the dashboard, login flow, settings page, mobile version, and three colour schemes. Sisinty tested it with "redesign India's Income Tax portal" and got a complete filing experience in 90 seconds: dashboard, refund status, ITR flow, and past returns view.

Product teams at Stripe and Coinbase use it for first-draft explorations. Figma Make handles mobile better but shows weaker systems thinking.

Prototyping

Make produces working prototypes with buttons that click, modals that open, and notifications that expand. You can hand these to stakeholders who interact with them like real apps. The integration into Figma is seamless. The trade-off: it's the slowest of the bunch, though still faster than manual work.

User-centric design

UX Pilot makes safe choices without being conservative. Strategic colour selections: green for good, orange for action needed, grey for neutral. Clear hierarchy guides the eye where it needs to go. It was fastest to generate and had the lowest wow factor, but produced the only design Sisinty would trust during tax season. The downside: sometimes too minimal, hiding complexity rather than organising it.

Design-to-code

v0 converts screenshots to production-ready React code in 30 seconds. The April 2025 v2 update made it precise enough that engineers use it regularly. The limitation: React and Tailwind only.

Microinteractions

Jitter builds butter-smooth microinteractions. Describe what you want: "Button scales to 0.95 on press, bounces back with elastic easing, ripples outward." It exports to Lottie or Rive and understands animation principles. Tell it "make this feel responsive" and it applies 40ms touch feedback, spring physics, and reduced motion preferences. The barrier: steep learning curve if you've never animated before.

Most improved

Make wins here, especially considering how heavily criticised it was at launch. Seamless integration into Figma made it part of most designers' existing workflows. Framer AI became more than a gimmick.

The bust

Adobe promised AI-powered UI design in Photoshop and Illustrator. What arrived: generative fill for mockups that look like 2015 Dribbble shots. Firefly generates designs that look good in screenshots but fall apart in use. No understanding of user flows. No design system awareness. The lesson: AI can't save a tool that's wrong for the job.

Overall winner

Figma Make takes the top spot not because it's the best at any single task, but because of where it lives. Every designer already uses Figma. Components and design systems live there. Team reviews and handoffs happen there. Teams at actual companies adopted Make because it enhanced existing workflows with negligible switching costs. UX Pilot makes better individual designs and Stitch is faster for exploration, but Make won the market.

Looking ahead

The tools generate interfaces well now. But they generate the same interface every time. Stitch, Make, and UX Pilot default to the same clean, boring, minimal SaaS aesthetic. None understand your brand yet. The 2026 challenge: brand-aware generative design that captures your design language and applies it consistently.