Maze.
Research,
sped up.

It is a Tuesday morning in a product team's Slack, somewhere, and a designer has already shipped a test. Not a feature - a test. By the time the standup begins, 42 strangers have clicked through her prototype, two have flagged the same checkout bug, and one has muttered, on camera, that the new button looks "like a tax form." This used to take six weeks.

WHO THEY ARE NOW

The quiet operating system of product discovery

Maze does not look like a company that has rewritten how software gets validated. Its homepage is calm. Its logo is round. Its colors are friendly. But underneath all of that politeness sits a research platform that more than 60,000 product teams now use to decide what to build, what to kill, and what to fix before anyone gets paged. Atlassian uses it. Braze uses it. Lowe's, Vimeo, Volvo, Inditex, Hitachi, Lenovo, PandaDoc and Safelite use it. It is, increasingly, the layer between a product idea and the people who will eventually have to live with it.

Maze is a Software-as-a-Service company, fully remote, with roughly 180 employees scattered across more than thirty countries. It runs on AWS, GCP, Ruby on Rails and a healthy amount of caffeine. Its annual revenue sits around $20 million. Its investors include Felicis, Emergence Capital, Amplify, Partech, Seedcamp, Atlassian Ventures, Zoom, and HubSpot Ventures - a list that, in any other industry, would look like a corporate party guest list. In SaaS, it just looks like consensus.

THE PROBLEM THEY SAW

User research had a release-cycle problem

For most of the software era, user research was treated like a fine wine. Expensive, occasional, and best brought out before a launch to impress visiting executives. It lived in dedicated departments at companies large enough to afford them, and almost nowhere else. Startups skipped it. Mid-market teams approximated it with hallway interviews and gut feel. Even at the enterprises that had researchers on staff, the cycle from "we should test this" to "here is what we learned" could stretch into weeks. By then the product had moved on, the designers had shipped, and the insight - if it ever arrived - showed up as a polite memo no one read.

The irony was thick. Software teams shortened every other feedback loop in their work. Continuous integration. Continuous delivery. Continuous deployment. They sliced bugs into branches and stories into points. And yet the one feedback loop that actually involved the humans paying for the software stayed stuck on a quarterly cadence. Research, the part most easily described as "talking to your users," somehow became the slowest thing in the building.

THE FOUNDERS' BET

Two French UX leads, one chat app, one accidental epiphany

Jonathan Widawski and Thomas Mary did not set out to build a research company. They were busy building Ping, a messaging app for gamers, and trying to validate it the old-fashioned way - by begging busy people to click through a Marvel prototype and tell them the truth. Two thousand of those busy people said yes. The duo had no good way to listen to that many at once. So they hacked one together.

The hack worked too well to throw away. In 2018, they spun it out and called it Maze. The bet was straightforward, even if it sounded a little impertinent at the time. Research, the founders argued, should not be the slowest part of building products. It should be the fastest. Anyone on a product team - designer, PM, engineer, the CEO who just had a hunch in the shower - should be able to run a usability test and get answers back the same day. Not in six weeks. Not after launch. Today.

It was an unfashionable position. Researchers, understandably, were nervous about a tool that promised to put their craft in the hands of amateurs. Maze's response was that this was not a takeover, it was a translation. The researchers still wrote the studies. The tool just made sure the studies actually got run.

THE PRODUCT

Everything a research team does, except slow

The Maze platform now stretches across the awkward middle of the product process. There is prototype testing for designs that have not yet shipped. Live website testing for products that have. Card sorting and tree testing for the people still arguing about information architecture. Surveys, screeners, and in-product prompts for the harder question of what users are doing on Tuesday afternoon. Reach, which launched alongside the Series B, runs the global panel - the actual humans who agree to be studied. Clips captures the moments worth replaying. Highlight reels stitch those moments into something a skeptical executive will actually watch.

And then there is the AI. Maze's AI Moderator runs interviews across time zones and languages with no human in the room, which sounds either futuristic or terrifying depending on which side of the calendar you sit on. Transcripts are automatic, summaries are automatic, themes are detected automatically. The researcher's job becomes setting the goals and judging the output, which is, arguably, what the job always should have been.

THE PROOF

A customer list that reads like an exit interview

The clearest signal that Maze had something the market wanted was who signed up to use it. Atlassian, who builds the tools that build the tools, became a customer and then an investor. Braze, the customer-engagement platform, runs studies on Maze regularly enough that it appears as a logo on the homepage. So do Lowe's, Vimeo, Volvo, Inditex, Hitachi, Lenovo, PandaDoc, NinjaOne, Iterable and Skillsoft. None of these companies needed another piece of software. They needed faster answers. Maze gave them those.

The reviews back this up in the language of B2B software, which is to say: Maze is a G2 Leader, a G2 High Performer four years running, and a GetApp Category Leader. Forbes added it to America's best startup employers for 2025, a fact made slightly funnier by the company having no office in America - or anywhere else. Maze is remote, fully. Its headquarters is wherever its laptops happen to be.

THE MISSION

Democratize research, even at the risk of researchers' eyebrows

The mission Maze keeps repeating - democratize product research everywhere - sounds like the kind of thing you write on a Notion page and forget. In Maze's case, it is the thing the product keeps measuring itself against. If a small team in Lagos can run the same usability study as a research department in Lisbon, mission served. If a PM in Sao Paulo can validate a flow before a sprint instead of after a launch, mission served. The company's own internal yardstick, judging by the cadence of its product roundups, is whether one more category of researcher gets to do their job faster this quarter.

There are competitors. UserTesting, which acquired UserZoom in 2023, remains the incumbent. User Interviews, Lookback, Dovetail, Sprig - all share the same neighborhood, often with overlapping features. Maze's pitch in that crowd is breadth: one platform that handles recruiting, testing and analysis, instead of three tools and a Zoom link. Whether the AI-first approach holds is the open question of 2026.

WHY IT MATTERS TOMORROW

The next product you use was probably tested on a Tuesday

Software is being shipped faster than ever, often by teams that did not exist eighteen months ago, often by AI systems that will not exist eighteen months from now. The bottleneck has shifted. It is not how fast you can build, it is how fast you can be sure you are building the right thing. That is the gap Maze is now sitting on top of - the gap between writing the code and earning the right to ship it. As AI builds more, research has to validate more, and validate it faster, or the whole thing devolves into a very expensive guessing game.

Maze is betting, with a quiet confidence that is unusual in its category, that user research will become as routine to a product team as code review. Not a department. Not a phase. A habit. If it is right, the unsexy back office of usability testing turns into the most leveraged tool in the modern product stack. If it is wrong, well, at least 60,000 teams will have had a really productive Tuesday.

Back to that designer, the one who shipped a test before standup. The 42 strangers who clicked through her prototype did not know they were participating in a small revolution. The checkout bug they flagged got fixed before lunch. The button that looked like a tax form was redrawn by mid-afternoon. The version that ships to a million users on Thursday will be the version those 42 people quietly approved. None of them will ever know. None of them needs to. That is the part Maze got right.