The Coliving AI Visibility Gap: What an AI Model Actually Knows About 103 Cities

Method note, read this first. This is a single-model recall study. I asked Claude (Opus 5, training data through May 2026) to name the coliving brands it associates with "best coliving in [city]" for each of the 103 cities Everything Coliving has published a city guide on. No web search. No retrieval. Pure model recall, which is the closest available proxy for what an LLM returns when it answers from parametric memory rather than browsing.

This is not the same as what ChatGPT, Gemini, or Perplexity return today. Those tools search live. But it is a clean measure of one thing that matters enormously: which coliving brands have accumulated enough durable, repeated, structured mention across the open web to be encoded in a model's weights at all. That is the deepest layer of AI visibility. If a brand is not there, it is dependent on winning a live retrieval race every single time.

The headline finding

I asked for ten brands per city, across 103 cities. That would be 1,030 names.

I could produce 1 city with a full ten. London.

86 of 103 cities could not produce five names.

10 cities produced zero.

The total real count across all 103 cities is roughly 280 brand mentions against a theoretical 1,030. That is a 27% fill rate. Three quarters of the coliving world does not exist in AI memory.

Brands recalled Cities Share 0 10 9.7% 1 to 3 66 64.1% 4 to 6 22 21.4% 7 to 9 4 3.9% 10 1 1.0%

The distribution is the story. Coliving talks about itself as a global asset class. In AI memory it is five cities and a long tail of silence.

Tier 1: cities with real depth (7+ brands)

Five cities. Two of them are in India.

• London (10) Folk Co-living, Mason & Fifth, Gravity Co-Living, Vonder, The Stay Club, Node, Dandi Living, Cohabs, AndSoul, The Collective.

• New York City (9) Common, Outpost Club, June Homes, Bungalow, SharedEasy, Ollie, Roomrs, Alcove, Node.

• Bengaluru (9) Colive, Stanza Living, Zolo, Your-Space, CoHo, Settl, Isthara, Homversity, NestAway.

• Mumbai (8) Stanza Living, Colive, Settl, Zolo, Your-Space, Hello World, Union Living, CoHo.

• Delhi NCR (8) Stanza Living, Zolo, Colive, CoHo, Settl, Your-Space, Placio, Oxfordcaps.

My read: India dominates this tier for a reason that has nothing to do with coliving quality. Indian operators built venture-funded, SEO-heavy, high-volume content operations and got covered relentlessly by Indian startup media. They are legible to a model. Whether they are actually coliving under a strict definition is a separate argument, and one I have made repeatedly. Much of this is PG stock with an app.

Tier 2: moderate depth (4 to 6 brands)

• Barcelona (6) Habyt, Aticco Living, Urban Campus, DoveVivo, The Social Hub, Node

• San Francisco (6) Common, Bungalow, HubHaus, Starcity, Canvas, Outsite

• Los Angeles (6) Common, Bungalow, PodShare, Tripalink, Treehouse, HubHaus

• Tokyo (6) Oakhouse, Social Apartment, Borderless House, Tokyo Sharehouse, ADDress, HafH

• Singapore (6) Hmlet, Cove, Coliwoo, lyf by Ascott, Figment, Casa Mia

• Paris (5) Colonies, Sharies, La Casa, The Babel Community, Kley

• Berlin (5) Habyt, Quarters, homefully, Stayery, Vonder

• Madrid (5) Urban Campus, Habyt, DoveVivo, The Social Hub, Livensa Living

• Lisbon (5) Outsite, Habyt, AMS Coliving, Draper Startup House, Selina

• Bali (5) Outsite, Selina, Tribal Bali, Roam, Livit Hub

• Hyderabad (5) Stanza Living, Zolo, Colive, Isthara, Settl

• Pune (5) Stanza Living, Zolo, Colive, Settl, Your-Space

• Amsterdam (4) The Social Hub, OurDomain, Vonder, Change

• Milan (4) DoveVivo, Joivy, Camplus, Habyt

• Washington DC (4) Common, Bungalow, Ollie, PadSplit

• Seattle (4) Common, Bungalow, Tripalink, PadSplit

• Miami (4) Common, Bungalow, Selina, Tripalink

• Shanghai (4) Ziroom, You+, Base Living, Mofang

• Hong Kong (4) Weave Living, Dash Living, Oootopia, Hmlet

• Jakarta (4) Cove, Rukita, Travelio, Mamikos

• Sydney (4) UKO, CDA Coliving, Scape, Hmlet Australia

• Chennai (4) Stanza Living, Zolo, Colive, Isthara

Tier 3: thin (1 to 3 brands)

Sixty-six cities. This is where the article gets its teeth. These are real coliving markets with real operators, and a model cannot name three of them.

• Three brands: Rotterdam (The Social Hub, Change, OurDomain) · Lyon (The Babel Community, Colonies, Kley) · Hamburg (Habyt, Stayery, SMARTments) · Munich (Habyt, SMARTments, The Fizz) · Stockholm (K9 Coliving, Allihoop, Colive Sweden) · Boston (Common, Bungalow, PadSplit) · Chicago (Common, Bungalow, PadSplit) · Austin (Common, Bungalow, PadSplit) · Denver (Common, Bungalow, PadSplit) · Toronto (Harrington Housing, Node, Sonder) · São Paulo (Yuca, Housi, Vitacon) · Dubai (Hive Coliv, Nest Dubai, Lodgd) · Seoul (Mangrove, Local Stitch, Come and Stay) · Beijing (You+, Ziroom, Harbour Apartment) · Kuala Lumpur (Hmlet, lyf, Cove) · Melbourne (UKO, Scape, Home Southbank) · Jaipur (Stanza Living, Zolo, Your-Space)

• Two brands: Manchester · Dublin · Brussels · Cologne · Frankfurt · Vienna · Copenhagen · Valencia · Porto · Rome · Florence · Warsaw · Krakow · Nashville · Houston · Montreal · Vancouver · Mexico City · Medellín · Buenos Aires · Tel Aviv · Cape Town · Johannesburg · Osaka · Bangkok · Chiang Mai · Kolkata · Ahmedabad · Chandigarh

• One brand: Birmingham · Edinburgh · Zurich · Oslo · Helsinki · Seville · Malaga · Athens · Prague · Istanbul · Bogotá · Lima · Lagos · Marrakech · Taipei · Hanoi · Ho Chi Minh City · Manila · Auckland · Kochi

Tier 4: zero

Ten cities where I cannot name a single coliving brand. Bristol · Geneva · Tallinn · Budapest · Bucharest · Belgrade · Tbilisi · Nairobi · Kathmandu · Colombo

Some of these are genuinely thin markets. Others are not. Tbilisi has been a nomad hub since 2020. Nairobi has active shared-living stock. Tallinn is a digital-first capital. The operators exist. They have simply produced nothing an AI system retained.

Four findings I would build the article around

1. Dead brands outrank living ones

The most uncomfortable result. These operators are shut, absorbed, or bankrupt, and they still surface strongly:

Brand Status Still recalled for The Collective Administration, 2021 London, top of recall Selina Bankruptcy, 2024 Lisbon, Athens, Medellín, Bali, Miami, Mexico City HubHaus Shut, 2020 San Francisco, Los Angeles Starcity Shut, 2022 San Francisco Quarters Insolvency, 2021 Berlin Danke Apartment Collapsed, 2021 Beijing NestAway Distressed sale, 2023 Bengaluru Oxfordcaps Wound down Delhi NCR Roam Shut Bali HmIet Absorbed by Habyt, 2022 Singapore, Hong Kong, Sydney, KL

A resident asking an AI for coliving in London in 2026 can be pointed at a company that collapsed five years ago. That is not a hypothetical risk. It is the current state.

The mechanism is obvious once you see it. These brands raised big, got covered heavily, then died. Coverage is what a model encodes. Nobody writes the follow-up article. The obituary never outranks the funding announcement.

2. Consolidation has flattened Europe

Habyt appears in Berlin, Barcelona, Madrid, Hamburg, Cologne, Frankfurt, Munich, Milan, Lisbon. DoveVivo, now Joivy, in Milan, Rome, Madrid, Barcelona. The Social Hub across the Netherlands, Spain and Italy.

Europe does not have a wide field of AI-visible operators. It has three or four roll-ups and a very long tail nobody can name. This maps exactly to the capital-side consolidation, which is the point I keep making. Demand was never the problem. The operating and capital side is where the sorting happened, and AI recall is now a lagging record of who survived the sorting.

3. The US names are platforms, not operators

Common, Bungalow and PadSplit account for most American recall outside NYC. Boston, Chicago, Austin, Denver and Seattle return an almost identical list. That is not five coliving markets. That is three national brands with SEO reach and no local competitor legible enough to displace them.

PadSplit is house sharing, not coliving. I have said this consistently and I am not softening it here. But it is winning the AI answer for "best coliving" in half a dozen American cities, which tells you the definitional problem is no longer just an industry-conference argument. It is being encoded.

4. Asia is the most under-indexed real market

Tokyo returns six. Singapore returns six. Then it drops off a cliff. Ho Chi Minh City, Hanoi, Manila and Taipei return one each. These are cities with substantial, functioning shared-living stock and genuine operators. The stock exists. The digital footprint does not.

If I were an operator anywhere in Southeast Asia, this is the cheapest competitive advantage available right now. The field is empty.

What this means for operators, the practical section

The gap between "we have a website" and "an AI system knows we exist" is the whole game, and most operators have not noticed it opened.

• What gets a brand into model memory. Repeated mention across independent, structured, crawlable sources. Directory listings with consistent NAP data. Industry publications. Comparison and city-guide content written by someone other than you. Structured data markup. Wikipedia-adjacent reference sources. Podcast and interview transcripts. The pattern is third-party corroboration at volume, not owned-channel volume.

• What does not. A beautiful website. Instagram. Paid ads. A booking engine. Every operator's current marketing budget, essentially.

• The 66-city opportunity. In any city in Tier 3 or Tier 4, one operator running a deliberate 12-month AI-visibility program can own the answer outright. Not compete for it. Own it. That window closes as the category matures, and I would put it at 18 to 24 months.

Source: single-model recall study, Claude Opus 5, training data through May 2026. City set drawn from the 103 unique cities covered in Everything Coliving's Coliving Spaces and City Guides category, September 2026.