Why we built natural-language search into Findr
Last spring, a photographer emailed us. She'd been clicking through filters for twenty minutes trying to find a studio with north-facing light, high ceilings, and a separate changing room. Twenty minutes for something she could describe in one sentence. That message sat with me for weeks.
The filter problem nobody talks about
Venue booking wasn't broken, exactly. People could find spaces. But the process felt clunky. A renter would start: location filter. Genre filter (studio, hall, meeting room). Then what? Size? Hourly rate? Whether the space has a kitchen? The filters worked, but they forced you to think in categories instead of describing what you actually needed.
The photographer's email was the third like it that month. Users were adapting their search to the software instead of the software adapting to them. That bothered me more than it probably should have.
We looked at our booking requests. The patterns were there. "I need somewhere quiet for podcast recording, close to the tube, under £30 an hour." "Meeting room for eight people, sofa would be nice, somewhere I can order lunch in." "Photography venue with blackout capacity and a mirror wall." These weren't filter combinations. They were sentences describing a need.
The decision to go conversational
Building natural-language search wasn't about being trendy. It was about matching how humans actually think. When you're looking for space, you don't mentally organize your needs into database columns. You say them.
We knew early on that this feature would live in our Plus tier, alongside advanced filters. The free tier remains simple: browse, filter by basics, send up to three booking requests a month. That keeps the barrier low. But for renters who book regularly, who know what they want, we wanted something faster.
The work started in earnest last autumn. Not just plugging something in; actually training it on real queries from our community. What do people actually search for when they describe a space? Lighting. Acoustics. Vibe. Proximity to parking. Whether the host is reliable (that one surprised us initially). We pulled request messages, looked at patterns, and built something that understood those words the way a real venue scout would.
What it actually does, without the jargon
Type something like "quiet studio with natural light and good wifi, south London" into the search bar. The system reads what you've written, understands you're looking for a creative space with specific conditions, and returns venues that match those criteria. No filter click required.
This works because we've mapped the language people use onto the data hosts provide when they list. A host might describe their space as having "floor-to-ceiling windows facing the garden." You might search "lots of natural light." The search understands both mean the same thing.
It's not magic. It's just thinking about search the way renters do, not the way databases do. You can still use filters if you prefer them. They're still there. But for most people booking regularly, typing your need beats clicking through six dropdown menus.
Why this matters for hosts too
Better search helps hosts as much as renters. A space that shows up for relevant searches gets seen by people who actually want to book it. Less tire-kicking, more qualified inquiries.
We've watched hosts obsess over their listing descriptions. Some write one paragraph. Others write paragraphs describing every detail, every fixture, what the light is like at different times of day. The detailed ones get more searches now. Their effort to describe the space properly actually pays off in visibility. That feedback loop is deliberate.
Hosts approve all bookings anyway, so natural-language search isn't letting wrong people through the door. It's just making sure the right people can find you. A host with a specialist music studio gets found by musicians. A gallery with high white walls gets found by photographers. The specificity works both ways.
The real advantage: you book faster
This is where the feature actually lives. Speed. A renter who knows what they want can describe it once and see matching results in seconds, then send a booking request. One request might be approved within hours. Compare that to the old path: filter, scroll, filter differently, scroll more, maybe not find it at all.
We built this into the Plus tier because we knew it would change how our regular users moved through the app. They'd save time. More bookings would happen. Hosts would get more qualified inquiries. The economics work for everyone.
The feature launched in January. Since then, Plus tier subscribers have booked 40% more spaces on average than Free tier users. Not because they're different people, but because the search matched their intent faster.
It's still a human marketplace
I want to be clear about something. The natural-language search is a tool for finding spaces faster. It's not deciding anything for you. You still message the host. They still approve or decline. You still see their calendar, their terms, their reviews. The booking is still between two people making a deal.
What changed is the first step. Instead of clicking filters and hoping, you describe what you need and see who's offering it. Then the human part happens. The conversation. The negotiation. The agreement. That's where real value lives, and that part is unchanged.
The photographer who started this whole thought process? She found a studio last month. North-facing light, high ceilings, separate changing room. Booked it in three minutes.
The question I keep coming back to is this: how many people are still clicking through filters to find something they could describe in one sentence? If that's you, it might be worth looking at whether the tool is actually fitting the way you work.