91% of respondents now use AI in their design work at least weekly, up from 54% in 2025—a 37-point jump year over year. 75% of surveyed designers use it daily.

AI has moved from "we use it somewhere in the process" to "we use it at every step." In 2025, the designers we surveyed were mostly using AI in the exploration and creation phases. Only 39% used it in delivery.

But today, every workflow we asked about shows meaningful AI adoption. The leading use cases are ideation, prototyping, and UI copy. And in our interviews, designers described continuous AI use across a project, rather than confining it to any one stage.

The biggest year-over-year movers include code generation, documentation, design QA, and developer handoff. These workflows sat closer to the edges of design practice a year ago. Most notably, designers in our interviews often mentioned their newfound ability to ship code and build high-fidelity prototypes. 50% of survey respondents say they’ve shipped code to production (read more about AI coding in Craft).

Designers we surveyed now use an average of 7 off-the-shelf AI tools, up from 3 last year—this is the case across respondents from different roles, company sizes, and geographies.

And this tally doesn’t take into account the full number of bespoke tools they build for themselves or internally built tools their company provides. We saw a major rise in these “internal” company tools—covered in more detail below.

Coding tools have found their place in the go-to design stack. In 2026, 76% of all respondents say they’ve used an AI coding tool like Claude Code, OpenAI Codex, Cursor, or GitHub Copilot. 85% have used these and/or an app builder like Lovable, Replit, or Bolt.

78% of respondents use Claude, compared to 65% for ChatGPT, which led in 2025. Gemini comes in third at 48%, and Perplexity has dropped from 34% to 13%.

65% of overall respondents are using Claude Code, which didn’t even make it onto our 2025 survey because it hadn’t launched to the public yet.

According to the UX Tools State of Prototyping survey, Figma remains the most-used design tool in 2026. But how designers use it may be changing. 

Some say Figma is their starting point for AI. They explore multiple directions in the canvas before going deep in code, or they create wireframes in Figma and pass them to AI. It remains a favorite place to ideate and collect inspiration, and respondents also pointed out that it’s still the best tool for team collaboration.

Other designers are using Figma as a “scalpel” or finishing tool in their AI-first design process. They prototype in a dev environment, then flip back to Figma for high-fidelity execution or precise tuning like perfecting radii and padding.

On the other end of the spectrum, we heard from code-first designers who rarely spend their time in Figma now. At Maze, designers work in Cursor from day zero. At Watershed, the design team has fully transitioned to working directly in the codebase.

Some designers are taking a middle path by using a newer generation of design tools like Paper, which are rebuilding the canvas in HTML and CSS. This allows them to continue using a canvas that connects directly to agentic tools, alongside other inputs like text prompts or code.

This tool adaptability is part of a broader AI-enabled shift that allows designers to customize their workflows—whether they want to start with code, mood boards, words, pen and paper, or an AI tool’s brainstormed idea. Ryan Mather, a designer at Anthropic, says, “The other week I painted an interface and shared it as input to my AI tool of choice!”

Just after we closed our 2026 survey, Anthropic launched Claude Design, an AI design workspace that generates complete prototypes, decks, one-pagers, and marketing assets. It’s accessible to non-designers like founders, PMs, and marketers who don't live in Figma the same way designers do.

We’re interested to see how this tool’s adoption will evolve and whether it becomes part of the standard designer’s toolkit in the next report.

80% of designers say reliable, high-quality output is what makes an AI tool stick. At the same time, 62% cite inconsistent or unreliable output as their biggest challenge when using AI for design work. We heard designers say the tools just “aren’t quite there” yet. 

But they also express how life-changing it would be if they were. Many believe that models will improve so rapidly that AI designs will soon pass art school standards and rival human craft.

Other major barriers to adoption include steep learning curves and poor integration with existing tools. At the company level, security and compliance concerns—along with budget constraints—continue to limit adoption, which may contribute to the proliferation of internally built design tools (see examples below).

Because their attention is spread across an ever-growing set of tools, designers are craving both a single, integrated platform that spans the full design process (in the vein of Figma or Sketch) and best-in-class tools for each step. They want seamless integration with existing systems and workflows. And they want to know that these tools will remain stable over time. 

In an environment where underlying models are evolving so quickly, that expectation may be difficult to meet.

Designers are building bespoke tools—first for themselves, then for their teammates—by encoding their judgment, taste, and design systems into workflow infrastructure. 

This is one of the most compelling shifts we’ve observed over the last year: designers as toolmakers.

The most sought-after designers today share this behavior. If designers were previously evaluated on their output, now they are evaluated on both their output and the workflows they build.

At the company level, designers are building shared tools that solve problems for other designers as well as engineering, PM, and marketing. 

The most AI-forward teams we spoke with—especially those at enterprises—have invested in infrastructure and promotional channels (like show-and-tells, internal marketplaces, and skill libraries) that make it easy for designers to evangelize and borrow home-grown tools. Read more in Teams.

74% of designers at enterprise companies (2,000+ employees) use internal tools, compared to just 26% at small organizations (up to 50 employees)—a 5x difference. At enterprises, these tools rank as the second-most-used general AI category, behind Claude.

Here are a few examples of how internal tools spread: 

• Internal open-sourcing: Anthropic has a separate GitHub repo that open-sources what team members build, making it easy to borrow from one anothers’ projects, and adapt them into their own bespoke tools.

• Playgrounds: Notion built a prototype playground—an internal, shared Next.js codebase where designers can create interactive product prototypes. Each designer works in their own folder, with a shared set of Notion-like design system components already available out of the box.

• Leaderboards: Stripe created an internal ranking of the most popular employee-built plugins and a spreadsheet of agents anyone can run.

Designers are feeling overall more confident in their AI toolstack compared to a year ago, but many are still exploring new options. Almost half of designers say they’re still searching for their go-to tools, while 37% say they’ve settled on a clear set of tools for most workflows. 

On average, designers who work at organizations with strong support for AI adoption report having higher confidence in knowing which AI tool to use for which design tasks.

The tools seem to change daily as capabilities evolve. Some designers describe this as a never-ending “molting process” that requires rethinking their entire workflow regularly. 

Many others report a growing workload, both to keep up with testing the latest tools and because expanded capabilities give them more possibilities of what to work on. Research from Harvard Business Review corroborates this: AI doesn’t reduce work, it intensifies it.