Finding housing on the internet feels more advanced than it actually is.
On paper, there are more places than ever to search: Facebook groups, Roomies, listing sites, group chats, Reddit, word of mouth, and random spreadsheets passed around by friends of friends.
But if you have ever seriously tried to find a room, a roommate, or someone to take over a lease, you know the experience still feels broken and time consuming.
By the numbers
“Trusting vibes” isn’t just a figure of speech. The scam problem in particular is well documented:
Where reported rental scams start
Source: Federal Trade Commission, “FTC Analysis Shows Consumers Have Lost Millions to Rental Scams”, Data Spotlight, Dec. 2025 (reports since 2020; platform breakdown for the 12 months ending June 2025). FBI figure from the Internet Crime Complaint Center’s 2024 IC3 Annual Report; that category also includes real estate investment and timeshare fraud and is not limited to rental scams. These are industry-wide figures, not Scout-specific data.
That is the problem we are building Scout to solve.
Scout is an AI housing agent that helps people find rooms, roommates, and room seekers without manually searching across fragmented platforms every day.
Tell Scout what you need once. Scout does the repetitive search, filtering, ranking, and compatibility work for you.
Most rental products treat housing search like a listings problem. You enter a city, a price range, maybe a few filters, and then you scroll.
That works fine when you are looking for a normal apartment. But shared housing is more complicated.
If you are looking for a room, the room itself has to fit you:
But the people also have to fit:
By the numbers: the person-fit problem
The traits people fight over aren’t guesswork. They show up consistently in what renters report after the fact.
What renters rate as the biggest roommate offenses
Sources: RENTCafé survey of 1,500+ US adults on roommate offenses; ApartmentGuide.com survey of 1,007 US adults who have lived with roommates, “Roommate Experiences Explored.” Industry-wide figures, not Scout-specific data.
And if you are trying to find someone to sign a lease with, there is a third problem:
The roommate search problem is both a room-finding problem and a person-finding problem.
Our product architecture starts from that insight. Scout’s first layer is a structured AI-powered onboarding interview with Nova, our agent, which creates a machine-readable “person-model” that powers listing ranking, compatibility simulation, and trust features downstream.
When you join Scout, you do not start by filling out a long static profile. You talk to Nova.
Nova asks about the practical constraints first:
Then Nova asks about the lifestyle details that usually get buried in awkward text conversations:
The goal is not to create a pretty profile page. The goal is to create a working model of what you need so Scout can act on your behalf.
That model then powers the rest of the product:
The first version of Scout helps users find rooms by taking messy housing supply and turning it into a ranked feed. Instead of manually checking every source, Scout ingests or imports listings, parses them into structured data, and ranks them against your Nova profile.
By the numbers: the search itself
The tedium isn’t just a feeling. It shows up directly in renters’ own search behavior.
How renters actually search (not mutually exclusive)
Source: National Apartment Association, citing an Apartments.com survey of 15,000+ U.S. adults planning to rent (May 2026), “Survey: Renters Rely on Rental Listing Sites”; search-duration figure from Apartments.com renter search data. Industry-wide figures, not Scout-specific data.
87% match: Bed-Stuy, NYC
$1,400/mo · Private room · Sept 1 move-in
Why Scout ranked it:
Watch: Pet policy unclear, so ask before contacting.
The point is not to show you every possible listing. The point is to show you the few listings worth your attention.
When you pass on a listing, Scout asks why:
That feedback matters. Scout is designed so every save, pass, and reason becomes signal that can improve future ranking. Saves make similar listing features more important, while pass reasons help Scout learn what users actually care about versus what they initially stated.
The onboarding interview with Nova gives Scout a starting model of what you want. But a stated preference and a real one are not always the same thing, and swiping is where the gap shows up.
Say you told Nova that pets were a dealbreaker, but you keep saving listings with cats in the photos and passing on cat-free ones for other reasons. That is a signal Nova did not have at onboarding. The next feed Nova builds for you should treat “no pets” as a soft preference, not a hard filter, without you ever having to go back and edit your profile.
You are not just training a feed. You are training your agent.
That is the distinction that matters here: the signal from your saves, passes, and pass-reasons does not just re-sort the current list, it updates the underlying person-model Nova uses everywhere: in how it explains listings, in how it ranks tomorrow’s feed, and eventually in how it evaluates roommate and lease partner compatibility. The more you swipe, the sharper Nova gets about what you actually want versus what you said you wanted on day one.
Where this stands today: seeker-to-seeker matching, the score, the match reasons, and the friction forecast below, is live in the app now. The seeker-to-listing and seeker-to-household legs described further down are still in design.
The more ambitious part of Scout is what we call A2A simulation: agent-to-agent compatibility.
This sounds technical, but the user-facing idea is simple: before you waste time DMing someone, Scout checks whether you are likely to be a good fit.
If you are looking for a roommate, your Scout agent already compares your profile with someone else’s Scout agent and surfaces the match, this part is live today. We are extending the same idea to two more relationships: if you are looking for a room, your Scout agent will compare you with a room poster or household; if you are posting a room, your Scout agent will surface seekers who already fit your price, timeline, house rules, and living preferences.
This gives Scout three matching relationships:
A normal platform might say: here are people looking for housing. For roommate matching, Scout already says: this person matches your budget, timeline, neighborhood, quiet weekday preference, and pet constraints, the main thing to clarify is guest expectations, and it says that before either of you has sent a message.
The point is not to replace human judgment. You still decide who to message, tour with, or live with. The point is to surface the right questions earlier.
Seeker-to-seeker matching runs as a compatibility simulation before either user knows the other exists, producing a confidence score, compatibility reasons, and a friction forecast about what might become an issue after move-in. We are extending the same simulation to listings and room posters next.
94% match
Why your agents matched you:
Scout’s forecast: Low friction overall. Watch: kitchen habits around week 4–5.
That last part matters. A lot of roommate products stop at “you are compatible.” Scout wants to go further: what specific issue might come up, and when should you talk about it?
Compatibility scores are easy to fake. Anyone can say “92% match.” The harder question is: was the prediction actually right?
That is why Scout is designed to follow up after a match. The long-term plan is to check in after 30, 60, and 90 days and ask whether the forecasted friction actually happened, how satisfied each person is, and whether they would live together again. We think of this as a calibration corpus: a dataset connecting pre-move-in compatibility forecasts to real post-move-in outcomes.
That feedback loop matters because housing is not just about finding something that looks good today. It is about finding something that still feels good after you have lived with the decision.
The internet is moving toward agents. Housing is a perfect early vertical for that shift because the search is fragmented, repetitive, emotional, and high-stakes.
People do not need another endless feed. They need an agent that can help answer:
That is what we are building with Scout.
We are starting focused. The early version of Scout is built around:
We are deliberately not automating leases, payments, or legal screening yet. That is a sequencing decision, not a ceiling. Trust has to be earned in order: get the search, the ranking, and the matching right first, on a system people can verify and correct, before extending the agent into the higher-stakes parts of the transaction.
Right now, Scout is built to do the exhausting first pass:
Then the human decides. As that foundation proves itself, we expect Scout to take on more of what happens after the match too, including leases, payments, and screening, with a human still able to step in at any point.
Scout is a housing-specific agent for the agentic internet. Instead of users manually searching Craigslist, Facebook groups, Reddit, Discord, and listing sites every day, Scout learns what they need and brings back rooms, roommates, and room seekers worth their attention.
The internet gave us more places to search. Scout is trying to make the search work for you.
Scout is live now in NYC and SF.
Start Scouting now