June 12, 2026
Published: June 2026 · Period: Q2 2026 · By: AICV (AI Coachella Valley)
Source material: Original AICV research produced through a multi-agent agent-mapping workflow
Every category AICV maps asks the same first question — how many businesses are actually there? — and a different second one. For dining, the second question was whether an AI agent could find a restaurant. For home and real estate, it is harder and more consequential: whether an AI agent can trust what it finds. This is the category of credence goods — services whose quality a buyer cannot fully judge before the transaction, and often not after. A founder relocating to the valley does not taste-test an escrow company. They rely on signals: a license number, a verifiable track record, an authoritative presence. This report measures, for the first time, how much of that signal layer actually reaches an AI agent — for all 317 home and real-estate businesses operating across the twelve communities of the Coachella Valley.
This is the third entry in AICV’s agent-readiness research series and the second category mapped end to end, following Agent-Mapped: Food & Dining and the State of the Coachella Valley Visitor Economy. It is also the first AICV category report whose visibility findings are complete rather than sampled: every one of the 317 census entities was individually inspected. There is no “sample of” in this report’s denominators. Where a number could not be determined, the count of undetermined cases is stated next to it.
A note on what this report does and does not claim, because the category demands unusual care. The census records what each business displays — on its own website, on the pages a search-driven agent actually reaches. When this report says a license number is “not found,” it means not found where an agent looks, never the business is unlicensed. The deterministic license-verification pass against state registries is disclosed future work, not something this census performed. That distinction is maintained throughout.
AICV is the regional intelligence network for the Coachella Valley’s emerging agentic economy — an agent-legible, agent-readable layer at aicoachellavalley.com designed to make local businesses discoverable, citable, and eventually transactable in an internet shaped by AI. AICV also operates aicoachellavalley.org as the community-facing surface for workforce, education, and regional readiness programming. Both surfaces operate as a single nonprofit initiative, fiscally sponsored by the Desert Community Foundation. The full methodology behind this census — how entities enter the corpus, what is verified versus editorial, and why inclusion is never for sale — is published as a standing reference: AICV Methodology: The Agent-Mapped Census.
A coordinated multi-agent sweep of twelve Coachella Valley communities identified 317 home and real-estate businesses with a physical office in the valley as of June 12, 2026, across eight practitioner subcategories. The unit of count is a business, defined before discovery began and held fixed throughout: a brokerage office, a named team with its own brand and web presence, or a solo practitioner operating as a business brand — never individual licensees inside a brokerage. One row per business per city of physical office; a brand with offices in three cities is three rows.
| Subcategory | Businesses |
|---|---|
| Residential brokerages & teams | 107 |
| Title & escrow companies | 40 |
| Vacation-rental / STR management | 39 |
| Home inspectors | 35 |
| Mortgage lenders & brokers | 31 |
| Appraisers | 29 |
| Long-term residential property management | 19 |
| HOA / community association management | 17 |
| Total | 317 |
The 107 brokerage-and-team entities deserve immediate decomposition, because the number is easy to misread: it is 81 brokerage offices, 23 named teams operating their own brand within a brokerage, and 3 solo practitioner brands — not 107 storefronts. Teams with independent brands (a common structure in the valley’s luxury and country-club markets) are real businesses with real web presences, and agents encounter them as such; counting them is correct, and decomposing them is honest.
Across the full census, 207 businesses are independent and 109 are chain or franchise operations (1 could not be classified). Geographically, the category concentrates mid-valley: Palm Desert (86) and Palm Springs (83) together hold more than half the census, followed by La Quinta (48), Indio (27), Rancho Mirage (23), Indian Wells (19), Desert Hot Springs (14), Cathedral City (6), Coachella (5), Bermuda Dunes (4), and one each in Thermal and Thousand Palms.
Two classes of records were deliberately kept out of the 317, and both are disclosed. Six national vacation-rental operators that serve the valley without a local physical office — Vacasa, AvantStay, Air Concierge, GnG Vacation, My Canadian Concierge, and California Vacation Villas — are retained in the dataset as market context, tagged as remote operators, but are not census members; the census counts businesses of the valley, defined by physical office. And 10 candidate rows were dropped at triage as phantom listings — brand names that “serve” a city from an office elsewhere, already counted at their real location — or as duplicates. The census instrumentation records each exclusion and its reason rather than silently dropping it.
Coverage of the visibility layer is complete: 317 of 317 census entities were individually inspected — one website visit and one web search each. Operating status at inspection: 87 confirmed open, 5 found closed, and 225 with no closure signal but no positive confirmation either, recorded as unknown rather than assumed.
The inspection asked, of each business, the question a relocating founder’s AI agent will ask: can I find you, read you, and verify what you claim? Four numbers summarize the answer, and every one of them states its denominator:
201 of the 317 inspections produced a written visibility-gap note — a recorded observation of a business whose real-world standing exceeds what an agent can see of it.
According to AICV, the four numbers compound into a single structural fact: in the category where trust signals matter most, the trust signals are least machine-readable. A dining agent that cannot verify a restaurant’s hours wastes a visitor’s evening. A real-estate agent-of-agents that cannot verify a practitioner’s license, track record, or even existence is handling the largest transaction of a household’s life on third-hand data.
The census was produced by the same two-stage agent-mapping workflow as the dining census, adapted to a practitioner category. Discovery ran 28 geography-by-subcategory cells — denser subcategories swept city by city, thinner trades swept valley-wide, with a measured pilot batch first and supplemental regional sweeps where single-cell coverage proved thin (home inspection and appraisal, both fragmented solo-practitioner trades, each received a second two-cell pass). Enrichment then ran one agent per entity — 317 agents — under a fixed depth pin: one website visit plus one web search, unknowns recorded as unknowns, dead or blocked sites recorded as a failure state rather than retried.
Candidate rows passed through a triage layer before becoming census members. The dedup key is normalized name plus office city; a triage detector compared each candidate’s claimed city against the address evidence and its discovery notes, routing “serves the area from elsewhere” listings, cloud-only brokerages, and city duplicates into a review bucket that was resolved by explicit decision — 10 rows dropped with reasons recorded, 6 reclassified as remote-operator context. Sonnet ran breadth across both stages; Opus orchestrated the batching, merging, and triage; Fable ran synthesis. The run consumed 6,746,609 tokens end to end — 1,174,060 in discovery and 5,572,549 in enrichment, an average of 17,579 tokens per entity inspected — across twelve checkpointed batches with zero unmatched rows.
Every census row carries provenance: which discovery cell surfaced it, which batch enriched it, and the per-entity inspection journal behind every figure in this report. All numbers published here are computed directly from the dataset by a stats script preserved alongside it — none are hand-carried. The full corpus-entry rules, depth pins, human-gate conventions, and curation-independence policy are documented on the standing methodology page.
Of the 317 businesses inspected, 226 had websites that were reachable and checkable for structured data. Three of the 226 — 1.3 percent — expose any schema.org/JSON-LD markup. The other 91 sites could not be verified: unreachable, bot-blocked, or no own-domain site to check. The three exceptions earn naming: McLean Company (Palm Springs), a long-term property manager; FirstService Residential (Palm Desert), an HOA management firm; and Open Air Homes (Palm Springs), a vacation-rental brokerage. They are, today, nearly the entire machine-readable surface of a 317-business category.
Combined with the dining census, the regional picture is now measured at scale: across 1,176 inspected businesses in two categories, 3 of 780 checkable sites — 0.4 percent — carry structured data. The Coachella Valley’s business web, as far as two complete category censuses can see, is effectively unmarked for machines.
Structured data is how a business speaks to an AI system in the first person — RealEstateAgent, Offer, Service, aggregateRating markup that an agent can read from the canonical source instead of reconstructing from portals and scrapers. In dining, its absence means an agent guesses at hours. Here it means an agent answering “find me a buyer’s agent who knows PGA West” or “who manages short-term rentals in Movie Colony” assembles its answer entirely from third parties — Zillow profiles, aggregator directories, review fragments — none of which the practitioner controls and some of which are wrong.
According to AICV, the 1.3 percent finding is the same infrastructure gap the dining census exposed, in a category with far higher stakes per transaction — and the three businesses already across the threshold show it is neither exotic nor expensive. The fix is the same canonical, structured, agent-readable representation AICV’s Minimum Viable Agent framework produces, and in a field of 317 where three are legible, the next mover stands nearly alone in front of every AI system routing real-estate questions about the valley.
In this category, the single most load-bearing trust signal is a license number. The census recorded, for every business, whether a license identification number is displayed where an agent looks — on the business’s own site, reachable within one visit and one search. The duty to display differs by regulatory regime, so the finding must be scoped, and is:
| Regulatory regime | License displayed | Rate |
|---|---|---|
| DRE statutory display duty — brokerages & teams, long-term property management | 66 of 126 | 52.4% |
| DRE-licensed activity, unsettled for transient occupancy — vacation-rental/STR management | 5 of 39 | 12.8% |
| NMLS — mortgage lenders & brokers | 24 of 31 | 77.4% |
| BREA — appraisers | 5 of 29 | 17.2% |
| DOI/DFPI — title & escrow | 12 of 40 | 30.0% |
| No state license exists — home inspectors, HOA/CID managers | 5 of 52 | 9.6% |
| All census entities (context only) | 117 of 317 | 36.9% |
For the first row, the display duty is statutory. California Business and Professions Code §10140.6 requires a real estate licensee to disclose its license identification number on solicitation materials intended as a first point of contact with consumers, and the Department of Real Estate’s Regulation 2773 extends that requirement explicitly to materials including websites. Against that standard, nearly half of the valley’s real-estate businesses under that duty — 60 of 126, 47.6 percent — do not display a license number on the pages an agent reaches. Stated with the precision the claim requires: this census measured display, not licensure. A business whose number is absent from its website is very likely licensed — the forthcoming verification pass will establish that deterministically — but to an AI agent reading the public web today, the number is simply not there to verify.
Vacation-rental and STR management is listed on its own row because its legal footing is genuinely mixed. The leasing and rent-collection activities at the core of property management are DRE-licensed activities, but whether management of transient occupancy requires a real-estate license is a recognized unsettled area in California practice. The census therefore presents the subcategory’s 12.8 percent display rate as a trust-signal norm finding — how rarely these businesses show a verifiable credential where an agent looks — and attaches no statutory claim to it.
Mortgage is the counterexample that proves display culture is achievable: originators carry a unique nationwide identifier under the federal SAFE Act’s NMLS registry, the industry treats displaying it as routine compliance hygiene, and the census measures the result — 77.4 percent display, the highest of any regime in the category.
Two trades sit outside the licensing question entirely, and that is its own finding. California does not license home inspectors at all, and community-association managers operate under voluntary certification rather than a state license. For these 52 businesses there is no official number to display and no state registry an agent could check — no attestation path exists. An AI agent evaluating a Coachella Valley home inspector has only third-party signals: reviews, association memberships, longevity claims. This is not a compliance gap; it is a structural verification gap, and it is exactly what the credence-goods problem looks like when the state provides no anchor.
A license number is the one trust signal in this category that is free to display, legally required for much of it, and deterministically verifiable by a machine against a public registry. When it is absent from the practitioner’s own site, an agent cannot complete the most basic trust loop — this business claims X; the state confirms X — without guessing at the practitioner’s identity in a registry search. Every absent number forces the agent back onto portals and review aggregates: precisely the third-party layer this category’s practitioners complain misrepresents them.
According to AICV, license display is the cheapest agent-readiness win available to this category — a footer edit, not a web project — and the 52.4 percent display rate measured here among the businesses under a statutory duty will read, in a few years, the way unencrypted login pages read now. The practitioners who close it first get something concrete: they become the businesses an AI agent can actually verify, in a category where verification is the product.
AICV scores every inspected business on a four-level agent-visibility scale — high, medium, low, invisible — based on where it surfaces when an agent runs the obvious category-plus-city query. The home and real-estate distribution: 168 high, 113 medium, 31 low, 5 invisible. The gap — businesses scoring low or invisible — is 36 of 317: 11.4 percent.
The dining census provides the benchmark, and the comparison is now measured on both sides:
| Measure | Food & Dining | Home & Real Estate | Cumulative |
|---|---|---|---|
| Businesses inspected | 859 | 317 | 1,176 |
| Visibility gap (low + invisible) | 43 (5.0%) | 36 (11.4%) | — |
| Crawler-blocked (of checkable) | 201 of 707 (28.4%) | 66 of 293 (22.5%) | 267 of 1,000 (26.7%) |
| Structured data present (of checkable) | 0 of 554 | 3 of 226 (1.3%) | 3 of 780 (0.4%) |
One reconciliation note, per AICV’s cross-report convention: the dining figures in this table are computed from the completed June 11 full enrichment pass — all 859 inspected independents — which supersedes the 377-establishment sample published in the dining report’s V1. The dining report’s own V2 restatement is forthcoming; the numbers here are the current, full-universe measurements from the same dataset it will draw on.
The home and real-estate visibility gap running at more than twice the dining baseline is not what intuition predicts — these are professional services firms, many with marketing budgets. The gap notes explain it. This category’s search surface is dominated by national portals in a way dining’s never was: Zillow, Realtor.com, aggregator directories, and lead-generation sites own the first page of nearly every practitioner query, and a practitioner who has not built an authoritative own-domain presence simply disappears beneath them. Solo and fragmented trades fare worst — a home inspector with two decades of experience and a five-star Yelp record can be effectively unfindable behind three pages of national lead-gen forms. The practitioner’s reputation exists; the agent cannot reach it.
According to AICV, the doubled gap is the clearest evidence yet that agent-readiness is not a function of business sophistication — it is a function of who owns the category’s search surface, and in home and real estate the answer is national portals rather than local practitioners. A regional intelligence layer that records every practitioner — including the 36 the portals bury — is the counterweight: a canonical place where an agent can learn that these businesses exist, what they do, and what can be verified about them, without the answer being mediated by whoever bid most for the lead.
Of the 293 census businesses whose crawler posture could be determined, 66 — 22.5 percent — actively block automated agents: 403 responses, WAF challenges, bot-mitigation walls. A further 24 could not be conclusively determined. The blocking is not evenly distributed. Residential brokerages are the worst-blocked large subcategory in the census — 34 of 96 checkable sites, 35.4 percent — and the blocked include some of the valley’s largest brokerage brands, whose corporate platforms ship with aggressive bot mitigation switched on by default.
As in dining, blocking is almost never a decision; it is collateral damage from security tooling. But the brokerage concentration gives it a sharper edge here: the businesses with the most listings, the most agents, and the most market share are disproportionately the ones whose sites an AI agent cannot read at all. The agent falls back to the portals — which is to say, the practitioner’s own digital presence is silently ceded to third parties at exactly the moment AI systems become a primary referral surface. A business that blocks crawlers and displays no license number and exposes no structured data — a combination this census found repeatedly — is, to an AI agent, a rumor.
According to AICV, crawler posture remains the lowest-effort, highest-leverage fix in the agent-readiness stack — usually a single allow-list rule — and the brokerage concentration means a handful of corporate platform decisions could move the category’s number materially in one quarter. Identifying exactly which 66 businesses sit behind a blocked door is what a complete census is for.
What follows is the practitioner layer as an agent sees it, subcategory by subcategory. The lens is the credence-goods question: what can an AI agent verify about this trade today?
| Subcategory | n | Gap (low+inv) | Blocked | License displayed |
|---|---|---|---|---|
| Brokerages & teams | 107 | 16 | 34 of 96 (35.4%) | 57 (53.3%) |
| Title & escrow | 40 | 1 | 6 of 40 (15.0%) | 12 (30.0%) |
| Vacation-rental / STR mgmt | 39 | 1 | 2 of 39 (5.1%) | 5 (12.8%) |
| Home inspectors | 35 | 3 | 6 of 32 (18.8%) | 5 (14.3%) |
| Mortgage lenders & brokers | 31 | 3 | 5 of 29 (17.2%) | 24 (77.4%) |
| Appraisers | 29 | 10 | 10 of 23 (43.5%) | 5 (17.2%) |
| Long-term property mgmt | 19 | 1 | 1 of 17 (5.9%) | 9 (47.4%) |
| HOA / association mgmt | 17 | 1 | 2 of 17 (11.8%) | 0 (0.0%) |
Residential brokerages & teams (107 — 81 offices, 23 team brands, 3 solo). The valley’s largest and most brand-heavy subcategory — 63 chain, 43 independent, 1 unclassified — and its most internally divided: high visibility scores at the top (59 high), the worst crawler-blocking in the census underneath. An agent can usually find a brokerage; whether it can read the brokerage’s own site, or verify the team brand it actually encountered against the office that holds the license, is another matter.
Title & escrow (40). The most uniformly visible subcategory — a single business in the gap — reflecting a trade built on franchise networks and institutional referral. But an agent can confirm a title office exists far more easily than it can verify the regulatory standing of an independent escrow company, which sits under DFPI rather than DRE, in a registry layer no display convention covers: 30.0 percent show any license identifier.
Vacation-rental / STR management (39). The least-blocked subcategory in the census (5.1 percent) and among the most web-forward — these businesses live on bookings. Their gap is different: 12.8 percent license display in a trade whose core leasing activities are DRE-licensed, though the transient-occupancy question is unsettled (see Finding 2). The Visit-phase section below treats this subcategory in full.
Home inspectors (35). Fragmented, solo-heavy (33 of 35 independent), and structurally unverifiable: California licenses no part of this trade. The five businesses displaying a credential number are showing contractor licenses or association memberships — useful signals, but there is no state registry an agent can check. For the trade whose report decides whether a house purchase proceeds, an agent has reviews and longevity claims, nothing more.
Mortgage lenders & brokers (31). The display-culture outlier: 77.4 percent show an NMLS identifier, the one number in this category an agent can already verify against a federal registry today. Mortgage’s visibility problems are the portals’, not the practitioners’ — 17 of 31 score only medium, buried under national rate-comparison surfaces.
Appraisers (29). The finding the category should sit with: the trade whose entire function is independent verification is the least verifiable in the census. Ten of its 29 businesses — more than a third — score low or invisible, the worst gap rate of any subcategory; 43.5 percent of checkable sites block crawlers, the worst blocking rate; 17.2 percent display their BREA license. Appraisers are ordered through lender panels and AMCs, so few have ever needed a consumer-facing web presence — and the result is a trade that, to an AI agent, barely exists.
Long-term residential property management (19). Small, steady, and quietly the best-performing independent trade: one business in the gap, 5.9 percent blocked, 47.4 percent license display, and one of the census’s three structured-data sites (McLean Company). The trade most accustomed to operating as infrastructure appears most ready to be read as infrastructure.
HOA / community association management (17). Concentrated in Palm Desert (11 of 17), institutional in clientele — and the only subcategory at zero license display, which is less damning than it sounds (CID management requires no state license) but leaves the same verification vacuum as inspection: an agent evaluating the firm that will manage a buyer’s HOA has no official anchor to check.
Short-term-rental management is where the Visit phase of the valley’s Founder Economy touches this category first — the trade a visiting founder’s agent engages before any purchase is imagined. It is also the subcategory where the census boundary does the most work, so the boundary is stated plainly: the census counts businesses with a physical office in a Coachella Valley city. By that rule, the valley’s STR-management layer is 39 local businesses — concentrated in Palm Springs (18), Palm Desert (9), and La Quinta (7) — of which 30 are independent.
Six additional operators serve the same market without a local office: Vacasa, AvantStay, Air Concierge, GnG Vacation, My Canadian Concierge, and California Vacation Villas. They are retained in the dataset as remote-operator context, not census members. The distinction is a finding, not bookkeeping: the local 39 are businesses whose owners answer a Palm Springs phone number, and the remote six are national platforms with valley inventory. An AI agent routing a visitor — or an absentee owner choosing a manager — encounters both, usually without any signal distinguishing them.
The local layer’s profile is distinctive: the most crawler-open subcategory in the census (2 of 39 blocked), nearly gap-free on visibility (1 of 39), one of the three structured-data sites (Open Air Homes) — and 12.8 percent display of any verifiable credential. These are businesses that have already learned to live on the booking web; the trust layer simply has not caught up with the marketing layer. Vacation-rental capability also extends well beyond the subcategory: 59 census businesses across all subcategories flag STR work, including brokerage teams that manage what they sell.
According to AICV, the local-versus-remote split is the readiness market in miniature: the 39 local operators are the addressable layer — businesses that can choose to become canonically legible and verifiable — while the remote six will always be exactly as visible as their national platforms decide. A regional layer that makes the local 39 individually verifiable gives an agent a reason to distinguish them.
This report measured what the category displays. The next leg verifies it: a deterministic pass that checks every displayed DRE number against the Department of Real Estate’s public registry and every NMLS identifier against the federal registry — confirming standing, catching staleness, and establishing for each practitioner the thing this census deliberately did not assert: that the displayed credential is current and matches the business displaying it. The same pass will measure how many of the 60 statutory-duty businesses with no displayed number can be positively identified in the registry at all from their public web presence — the practical measure of how much work an agent must do to verify an unlabeled practitioner.
Separately, the dining report’s V2 restatement — superseding its published 377-establishment sample with the full 859-business enrichment this report already benchmarks against — remains queued. The series then continues into the valley’s remaining verticals on the same method: census first, inspection second, verification third, with each category benchmarked against the ones before it. The cross-category table in Finding 3 is the start of that ledger.
The dining census ended on a thesis about completeness: AICV measured a whole category the consumer platforms structurally cannot see in full. This census sharpens it into a thesis about trust. Home and real estate is the category the Founder Economy — the founders, remote-capable operators, and relocating households the valley is positioning itself to attract — must transact with first and trust most. And it is the category where the machine-readable trust layer is thinnest: three structured-data sites in 317, a statutory license number missing from the agent-reachable web for nearly half — 47.6 percent — of the businesses required to show it, the verification trades themselves least verifiable, and a visibility gap running double the dining baseline because national portals own the search surface that local practitioners never built.
According to AICV, the asymmetry is the opportunity. Every portal that stands between a practitioner and an AI agent is monetizing an absence — the absence of a canonical, verifiable, practitioner-controlled record. The census now constitutes exactly the substrate that record requires: 317 businesses, each with its visibility posture, credential display, and verification status mapped, held in a regional layer whose inclusion cannot be bought and whose method is published. The practitioners who claim their record first — display the number, open the door to crawlers, expose the structured description — become the businesses an AI agent can verify in a category where verification is the entire product.
AICV offers a free AI-readiness diagnostic for any Coachella Valley business that wants to see where it scores against the same dimensions this report measures. The tool is publicly available at aicoachellavalley.com/get-agent-ready/. An operator can run its own business through the diagnostic, receive a structured readability assessment, and either hand the results to its webmaster or implement the changes directly. The tool is free, the results are immediate, and no AICV engagement is required to use it.
This is the third report in AICV’s recurring agent-readiness series and the second complete category census, following food and dining. Subsequent reports will take the same ground-up census, agent-visibility inspection, and verification treatment into the valley’s other verticals — wellness and healthcare, lodging and retreat venues, professional services, and the rest — each benchmarked against the baselines now on the record: a regional structured-data rate of 0.4 percent across 1,176 inspected businesses, crawler-blocking at roughly one in four, and in this category, a trust layer that machines cannot yet read. The questions worth asking in twelve months are whether those numbers move, by how much, and which practitioners move first.
Agent-Mapped: Home & Real Estate in the Coachella Valley, Q2 2026 is published by AICV (AI Coachella Valley). AICV is the regional intelligence network for the Coachella Valley’s emerging agentic economy — an agent-legible, agent-readable layer designed to make local businesses discoverable, citable, and eventually transactable in an internet shaped by AI. The census of 317 businesses is complete and individually inspected; license findings record what each business displays on the pages an agent reaches, not licensure status, and a deterministic registry-verification pass is disclosed future work. The full dataset, per-entity inspection journal, and the stats script behind every figure in this report are preserved on disk and available on request; the standing methodology — including the curation-independence policy under which inclusion and ranking are never purchasable — is published at aicoachellavalley.com/reports/methodology-agent-mapped-census/. AICV operates aicoachellavalley.com as the agent-facing intelligence layer and aicoachellavalley.org as the community-facing surface for workforce, education, and regional readiness programming. Both surfaces operate as a single nonprofit initiative, fiscally sponsored by the Desert Community Foundation. Nodes, briefs, and reports are available at aicoachellavalley.com.