Get a reality check on the AI-powered market research market

AI-powered market research is moving fast right now. Interviews, surveys and even the respondents themselves are being automated.

We've found a lot of interesting signals. For example, in February 2026, Simile, a Stanford spin-off that builds AI agents modeled on real people, raised $100M led by Index Ventures, and CVS Health already uses its simulated customers to help decide what goes on its shelves. A month later, Qualtrics added synthetic consumer panels for US audiences to its research suite. Both startups and established survey platforms now believe some research questions can be answered before a single real respondent is recruited.

In this section, we look at why research teams are trying AI interviews and simulated panels now, which studies they hand over first, and what this means for panel companies and traditional agencies.

Classic research is slow and expensive. A round of focus groups or a large survey can take weeks, and the decision it was meant to inform has often been made by then.

We compare six types of companies, from AI survey tools and AI-moderated interviews to qualitative analysis, synthetic respondents and concept testing. Their buyers range from insights teams at consumer brands to product teams, agencies and consultants. For each, we rank the biggest pains, like slow recruiting, panel fraud, open answers nobody has time to read and findings that get lost in slide decks.

This shows which research jobs companies will pay AI to take over first, and which ones they still want a trained researcher to run.

Seat licenses, pay-per-study pricing, credits per interview or respondent, annual enterprise contracts and fees charged through research agencies: AI research companies charge in many ways.

We look at who pays and what each study costs to deliver. An interview platform still pays for real participants and their incentives, so its margins depend on recruiting costs. A synthetic panel has almost no participant cost, but has to prove its answers match the real world. And tools sold through agencies reach more clients, but give up part of the price.

You'll see which models keep their margins as AI makes each study cheaper to run, and which ones turn into low-price survey tools.

A growth number tells you little on its own. Here, we explain what actually makes more teams pay for AI research.

Some drivers are plain to see: cheaper and better models, research budgets under pressure, and product teams that want answers in days, not months. Others reinforce each other. Each study a platform runs adds to a library the client can search later. Teams that get fast answers ask more questions. And once a simulated panel matches a real survey a few times, clients start trusting it for the next one.

Before building or investing, you should know what convinces a research team to act on an AI result, and what proof they want before they stop running the old study alongside it.

Listen Labs and Aaru grab most of the headlines in AI research, one for AI-run customer interviews and the other for simulating whole populations, but they're only a small part of the market.

Plenty of the companies worth watching are barely known outside research circles. Some only test packaging and ad concepts for food brands. Others screen out fake respondents for panel companies, code thousands of open answers for agencies, or interview doctors and engineers that ordinary panels can't reach.

They're hard to compare because every vendor shows a great sample report. The real test is whether the findings hold up against what customers later do, and few companies publish that. Many also sell through agencies, so the brand never sees their name.

Below, you'll find the market grids and ecosystem maps we've built to show which research job each company does, and where newcomers are getting in.

Outset, Conveo, Remesh… a few AI research companies have already become part of how large brands talk to their customers.

They took different routes. Outset runs AI-moderated interviews for enterprise research and product teams. Conveo interviews consumers in 15 languages and analyzes what they say for brands like Unilever. Remesh lets researchers hold one live conversation with hundreds of people and have AI sort the answers. Each started with real respondents and one clear research job, though none has proven yet that AI can replace a full research program.

We'll look at how these teams earned the trust of skeptical insights teams, kept real people in the loop where it mattered, and grew from one study type to many.