Shane JohnstonResearcher, FigmaIn 2024, we kicked off a study to understand how AI would reshape design and product development. Now, AI is helping us run it.
Share How do you actually measure the impact of AI on design?
Illustrations by Zoé Maghamès Peters
When AI tools and products started gaining momentum a few years ago, we knew we wanted to track their impact on the design industry, from ways of working to product development strategy. We had questions big and small: How much time would AI really save? How would AI change the way teams collaborate, and how did people feel about it? And what tools could help us measure AI’s impact holistically?
New technologies like AI don't transform industries all at once; the effects compound over time. So we set out on a multi-year research project AI used to be a solo act. According to our research, it’s not anymore.Figma's 2026 AI report: Can AI help us collaborate better?
Here's what we built, and what we learned about using AI to conduct research at scale.
Download Figma’s 2026 AI report to explore the trends we’ve been tracking.
A single scale for a moving target
First, we decided to create a sentiment index, a single scale that reflects how a group of people feel about something. The best-known example is the University of Michigan's consumer sentiment index, which tracks how optimistic or pessimistic people are about the economy. It doesn’t measure what the economy is actually doing; it captures feelings of confidence or anxiety and how it may influence behaviors like spending habits. That’s exactly the read we wanted on AI: not just what AI can do, but how the work is changing, how people feel about it, and what it means for the future of design.
We ask participants what they’ve experienced in the past 12 months, and what they expect to experience in the next 12. Comparing predictions versus reality tells us whether AI is keeping up with expectations, outpacing them, or falling short.
In 2026, every dimension exceeded last year's expectations by 10+ index points—collaboration nearly doubled from 32 to 58, while personal productivity, projects, and products climbed into the high 60s and 70s from the 40s.
To build Figma’s AI impact index, we asked designers, developers, and PMs questions like “Over the past 12 months, to what degree has AI affected your personal productivity and workflows?” and “Over the next 12 months, to what degree do you expect AI to impact how you collaborate with others?” We mapped answers on a scale from 0 (no impact) to 100 (transformational). A score around 50 means AI is becoming part of someone's job without fully reshaping it; scores climbing toward 100 mean AI is fundamentally changing their day-to-day.
As with any new technology, AI impacts different aspects of work unevenly—for example, an individual’s productivity shifts before a whole team’s does. That’s why we shaped our questions around six dimensions of product builders’ jobs. We’ve used the same questions for the past three years, so we can track trends in each area:
The survey focuses on six different aspects of product builders’ lives at work, inspired by socio-technical systems theory and research on digital transformation.
- Their organization’s goals: If AI changes what companies prioritize and where they place bets
- Their productivity and workflows: How people use AI for daily work—like exploration, prototyping, and coding
- How they collaborate with others: AI’s influence on how teams work together
- The types of projects they work on: How AI plays a role in what people are actually building
- Their company’s products and services: Whether companies are selling AI-powered products
- Their tools: How AI changes the tools people use
Beyond the index: AI interviews at scale
Figma’s AI impact index tells us what AI has changed, and by how much. But these numbers don’t tell the full story. This year, we wanted to hear firsthand from the people actually experiencing these changes. To do that at scale, we tried an AI interview tool.
An AI moderator leads live conversations with participants, asking clarifying questions and adapting to context in real time to get the most thorough answers. We ran 639 AI-moderated interviews, covering everything from the specific ways AI has transformed people’s work, to their concerns about the future, to how their organizations are responding. We collected more interviews than would have been humanly possible in the same amount of time, so we wanted to use AI to help analyze the correspondingly high volume of data, too.
2026 interviews by the numbers639 transcripts
2 million words
40,701 codes
11,846 total themes
553 recurring themes across participants
But off-the-shelf research analysis tools we’ve tried are opaque. If we can’t see how a tool is interpreting the data, there’s no way to check its work. So we built one that gave us visibility into every step: an AI-powered version of a thematic analysis, a standard research method that involves reviewing every transcript, labeling ideas (“codes”) within each interview, and synthesizing them into broader themes. Then, we could review any theme—for example, “AI increased iteration speed but introduced quality concerns”—and easily see the 47 participants who mentioned it, read their exact words, and trace each one back to the original interview transcripts.
MIT economist David Autor coined the term “expertise re-intermediation”: AI doesn't eliminate the need for expert judgment, but makes that judgment scalable.
The AI analysis wasn’t just faster; it was more comprehensive than our team could achieve alone. When a human is reviewing hundreds of interviews by hand, they might code the 50th interview differently than the first. Fatigue sets in, interpretations drift, and patterns noticed early on influence how they read later transcripts. But our tool used a consistent analytical lens on every interview, every time.
At the same time, we couldn’t just punt the process to AI; we needed to design a system that enforced rigor at every step. We had to give it clear context: what methods to use, the level of granularity we wanted, how to handle conflicting themes, and how to trace themes back to each transcript. AI didn’t replace our expertise as researchers, but it gave us the infrastructure to apply that expertise again and again.
Building a cross-functional source of truth
The thematic analysis tool solved one problem, but it created another: Now, we had tens of thousands of data points across hundreds of interview themes. All of that lived in notebooks and text files that would be difficult for a non-researcher to navigate. But reporting is a team sport. The people behind Figma’s 2026 AI report spanned researchers, writers, marketers, and analysts—and they all needed a shared source of truth without waiting on me to pull numbers or export charts.
We deployed the dashboard site using Payload CMS, which gave us version control, access management, and the ability to easily share it.
So this year, we coded an internal dashboard that pulled together the full dataset: three years of survey results, interview analyses, and interactive charts showing trends, breakouts by role and region, and deep dives into each index dimension. To build it, we converted the raw data analysis into a self-contained HTML document, then used Figma’s MCP server to iterate a few times. We designed the layout and interaction patterns in Figma, generated the initial code structure, refined the output in Figma to nail the styling, and tweaked until it worked.
The dashboard was more than a place the data could live; it was a shared tool anyone across the team could use so we could work together better. Instead of everyone forming independent mental models from scattered emails and slide decks, we all worked from the same source.
Figma’s AI impact index itself tells us AI is changing design work fast—reaching 62 index points out of 100 this year, nearly double what it was in 2024. But even before we finished the analysis, AI’s impact was visible to us; the index was how we put a number on it.
To dig deeper into the index, read Figma’s 2026 AI report: The new era of multiplayer design.
Shane Johnston is a researcher at Figma where he explores how AI is reshaping design and product development. Previously, Shane's research focused on the future of work at Slack's Future Forum, communities at Twitter, and consumer psychology at Procter & Gamble's cognitive science group.
Figma's 2026 AI report: Can AI help us collaborate better?
AI fluency isn't the finish line
Figma's 2025 AI report: Perspectives from designers and developers