Sonder's Anti-AI Bet: Friction as a Feature, Not a Bug

A new London dating app is deliberately making sign-up tedious—and people are queuing up anyway. Sonder has built a 6,500-strong user base by forcing singles to spend up to two hours creating freeform 'mood boards' instead of answering quick prompts, inverting a decade of industry optimisation around frictionless onboarding. The entire growth has come through word-of-mouth, according to founders who previously built AI systems at Meta and are now betting against algorithmic matching entirely.

Friction as filtering

The core thesis is simple: make the entry barrier high enough, and you filter for people who actually want to be there. Sonder's sign-up process requires users to create what amounts to a digital collage—a Pinterest board crossed with a personal essay, with no character limits, no multiple-choice questions, and no algorithmic hand-holding. Users can upload images, write freeform text, and link to music, articles, or anything else that 'represents them'. The company reports that profiles take between 45 minutes and two hours to complete.

That's not a bug. It's the entire product strategy.

Create a free account

Unlock unlimited access and get the weekly briefing delivered to your inbox.

Register free

No spam. No password. We'll send a one-time link to confirm your email.

The time investment screens out casual browsers, bots, and anyone not genuinely committed to finding something real

Compare this to the industry standard. Match Group (MTCH) has spent years optimising Tinder's onboarding flow to reduce drop-off. Bumble (BMBL) guides users through structured prompts designed to maximise profile completion rates whilst minimising cognitive load. Hinge famously built its brand on prompts that 'designed to be deleted'—which in practice means designed to be answered quickly. The entire sector has moved towards reducing friction at every turn.

Sonder is betting the opposite: that friction itself is the filter. Whether that's true at scale remains untested. What's notable is that the hypothesis has resonated enough to build a five-figure user base in a single city without paid marketing—a claim DII cannot independently verify, but which the founders disclosed to TechCrunch.

The hybrid model

Sonder isn't just an app. The company runs regular in-person events in London—dinners, parties, group activities—that function as both community-building and acquisition channel. Members meet offline, talk about the app, invite friends. The model looks more like a members' club than a platform.

This mirrors a broader shift among younger operators trying to solve the same problem: how do you build trust and authenticity when the core mechanic is still swiping through strangers? Thursday pivoted hard into events after realising its one-day-a-week gimmick wasn't defensible as a pure-play app. Feeld has experimented with IRL gatherings for its kink-curious community. Even Match Group has tested offline experiences through Archer, its London social club.

What's different here is that Sonder launched with the hybrid model baked in. The app and the events aren't separate revenue streams or bolt-on features—they're the same product. Users download the app because someone they met at a Sonder event told them to. They attend events because someone they matched with on the app invited them. The loop is the proposition.

Whether this works outside London is unclear. The model depends on density, both geographic and social. You need enough people in a small enough area to make recurring events viable, and you need enough social overlap for word-of-mouth to compound. That's easier in Hackney than in Houston. For those curious about navigating the London social scene, the city's unique density creates both challenges and opportunities for connection.

The AI engineer who built an anti-AI app

There's a certain irony in the fact that Virkus spent years at Meta working on AI systems before co-founding a dating app that explicitly rejects AI assistance. Sonder doesn't use algorithmic matching. It doesn't suggest prompts. It doesn't auto-generate icebreakers or optimise your profile for engagement. The founders describe it as 'anti-algorithm', which in practice means: here's a blank page, fill it yourself, good luck.

A former Meta AI engineer has built a dating app that explicitly rejects AI assistance—no algorithmic matching, no suggested prompts, no optimisation

This positions Sonder squarely against the current trajectory of the industry. Match Group has bet heavily on AI-powered features across its portfolio—Tinder's photo selection tool, Hinge's voice prompts, OkCupid's compatibility scoring. Bumble launched an AI photo analyser and has publicly discussed using machine learning to surface better matches. Grindr (GRND) has tested AI chat suggestions. The entire sector sees AI as the answer to engagement and retention problems.

Sonder's founders see it as the problem. Or at least, they see it as something Gen Z increasingly associates with inauthenticity. Whether that's a defensible long-term position or a cultural moment that passes in 18 months is the question investors will ask if Sonder ever tries to raise serious capital.

What this means for the rest of the market

If Sonder's model proves durable, it suggests a wedge in the market that the big platforms can't easily serve. Match Group can't make Tinder cumbersome without tanking engagement metrics. Bumble can't remove prompts without breaking the user experience it's spent years refining. The economics of public-market dating companies depend on scale and efficiency. A high-friction, slow-growth, events-heavy model doesn't map onto that playbook.

That doesn't mean it's irrelevant. It means it's a niche. The question is how large that niche is. If dating app fatigue is real—and the data from Pew, Axios, and MTCH's own user surveys suggests it is—then there's room for products that solve for trust and authenticity even at the cost of growth. But those products will likely stay subscale, or get acquired by larger platforms looking to hedge their bets.

The other possibility is that Sonder is just early. That the industry's decade-long march towards frictionlessness has overshot, and that operators who reintroduce friction—selectively, thoughtfully—will outperform. That would be a meaningful shift, and one worth tracking. The app's approach embraces a deliberately annoying sign-up process that completely abandons standard dating app conventions.

For now, Sonder is a London curiosity with 6,500 users and a good TechCrunch story. Whether it's a signal or noise depends entirely on what happens in the next 12 months. If growth stalls, it's noise. If it compounds without paid acquisition, it's a signal the rest of the industry should take seriously.

D