Investor Brief · 2026

The first AI brokerage in NYC. Patent-protected.

Built by a thirteen-year licensed luxury operator. A patent-protected software business inside a $100 billion commission pool — designed to do to residential real estate what Stripe did to payments.

A $100B industry running on 1990s code

They tried to replace the agent.
We replace the agent's busywork.

Residential real estate generates over $100B in annual US commissions on tech that hasn't fundamentally changed since the listings database. Compass burned $1.5B building a better CRM. Zillow lost $880M in a single quarter trying to buy houses. Both failed for the same reason.

Compass · Elliman · Redfin
  • Headcount businesses marketed as technology
  • 30,000-agent operations with −13% margins
  • 80% of revenue paid out in agent splits
  • No proprietary data, no data graph, no IP
  • Valued at less than 0.5× revenue
  • Cannot cannibalize their own agents to automate
Signal Homes · AI
  • Software business with brokerage compliance gating
  • Skeletal team, AI-native workflow, software margins
  • Automation captures the margin the legacy shops bleed
  • Patent-protected: 283 pages, 16 subsystems claimed
  • Shadow inventory graph: data nobody else can have
  • Built to be acquired or to acquire — not to plateau

The category is wide open. No AI-native incumbent exists in residential brokerage. Whoever ships first sets the default — like Stripe did for payments.

Why now

Three forces. One inflection.

Signal Homes isn't a clever idea — it's the inevitable consequence of three independent shifts that all crossed threshold in the last eighteen months. Miss this window and the category is built by someone else.

— 01 · Technical

LLMs crossed the line.

Large language models can finally hold a real conversation about a property without hallucinating — gated by deterministic compliance state machines (claimed in our patent) that make AI brokerage legally possible for the first time.

— 02 · Regulatory

The NAR settlement.

The 2024 NAR antitrust settlement unbundled buyer-side commissions for the first time in American real estate history. Buyers must now choose and pay for their representation explicitly — the door to a buyer-first platform is open.

— 03 · Market

NY luxury inflection.

REBNY's RLS rules and pending NY State legislation are reshaping off-market disclosure in the highest-margin US real estate market. The shop that builds for the new rules wins the next decade.

What we are

Three businesses. One platform. One patent.

Most startups bet on one revenue line. Signal Homes is structured so each line de-risks the others — and any one of them is fundable on its own.

— B2C · Direct revenue

Hudson — the AI broker

For buyers. NYC luxury wedge. Off-market inventory the search sites literally can't have. Negotiation trained on FBI tactical empathy. Customer pays nothing; we earn buyer-side commission, now explicit under the NAR settlement.

— B2B · Recurring SaaS

The multi-tenant layer

The platform sold back to brokerages as licensed infrastructure. Explicitly claimed in the patent. 110,000 US brokerages. At 1% penetration and $30K ARR per tenant: $33M ARR at software margins.

— Data · Moat

The shadow graph

A continuously-updated off-market property-and-owner graph with calibrated likely-seller scores. The most valuable proprietary dataset in residential real estate — and it compounds with every observation.

Hudson · the consumer product

Zillow shows you listings.
We show you the truth.

Hudson learns your taste through a conversation, hunts inventory across listed and off-market sources, and walks you through the truth behind the photos — light, board friction, sponsor history, building risk — that no search site can tell you.

StreetEasy · Zillow · Compass
  • Beds, baths, price, square footage
  • The same public listings everyone scrolls
  • Listing photos staged for sale, not for truth
  • No idea which board approves you
  • Zero context on the sponsor or the building
  • You scroll. They sell ad placements.
Signal Homes · AI
  • Real light: morning sun, afternoon, airshaft, southern exposure
  • Prewar detail: moldings, plaster, oak vs. parquet, fireplaces
  • Honest photos: AI flags wide-angle distortion and virtual staging
  • Board intel: approval friction, financials, recent litigation
  • Sponsor history: track record, comp concessions, refi signals
  • Off-market: owners likely to sell before they list

Every interaction sharpens the shadow graph. Every visit produces ground-truth observations that the search sites — by structure — cannot have. The data compounds. The moat thickens.

The defense

Eight jewels. One filing.

Our provisional patent claims sixteen interoperating subsystems across 283 pages. Below are the eight that matter most to the moat — each a specific computer-implemented improvement that cannot be performed by human judgment alone. Each independently claimable.

— JEWEL 01

The taste vector

High-dimensional preference modeling across five interaction signals — swipe, paired-comparison, dwell, conversation, mood — with signal-specific decaying learning rates and a drift-triggered soft reset. Spotify for homes, with claims behind it.

— JEWEL 02

The shadow inventory graph

A continuously updated graph of properties, units, buildings, owners, and ownership clusters with calibrated listing-probability and likely-seller scores. Zillow shows what's listed. We show who is about to list.

— JEWEL 03

Regulatory-threshold handoff

A deterministic finite automaton with cryptographic continuity tokens that detects when an AI conversation has crossed into regulated brokerage activity and routes to a licensed human. The only legally defensible way to run AI brokerage at scale.

— JEWEL 04

Ground-truth ingestion

Geofence-validated, privacy-processed visit observations bound to building-and-unit identifiers, aggregated across visitors with robust statistics. The data Zillow cannot have, because they have no agents in the buildings.

— JEWEL 05

Presentation-deviation index

Computer vision that compares listing-supplied photos against user-contributed photos along luminance, color cast, focal length, and sharpness. "The photos look better than reality" becomes a quantitative score that flows into match, valuation, and recommendation.

— JEWEL 06

Citation-enforced RAG · URL guard

Every assertion the platform makes is gated by a per-assertion citation validator. Fabricated URLs are blocked by a two-gate validator. Zillow's AI hallucinates the board approval policy. Ours can't.

— JEWEL 07

Tactical-empathy negotiation

A real-property-specific concession ladder with credible anchor, diminishing increments, non-round final offer, phantom-offer detection scored against eight indicators, and counterparty profiling bracketed by protected-class guards. Chris Voss, codified.

— JEWEL 08

Compliance-gated outreach

Cryptographic human-approval transmission tokens on every cold message. Protected-class input and output guards bracket every inference. Brokerages currently lose deals — and get sued — over this. We turn compliance into a technical guardrail.

The moat

Thiel's four conditions — all four present.

Peter Thiel's monopoly test: proprietary technology, network effects, economies of scale, and brand. Most startups manage one. Signal Homes is built to satisfy all four simultaneously.

Proprietary technology

Provisional patent filed May 2026. Sixteen interoperating subsystems. Non-provisional filed within twelve months. The compliance state machine alone is an eighteen-month engineering chokepoint.

Network effects

Every visit produces ground-truth observations. Every observation refines the shadow graph. The graph cannot be replicated retroactively — competitors arriving later face a dataset deficit that grows daily.

Economies of scale

Software margin replaces agent overhead. Compass spends 80%+ of revenue on agent splits. Our marginal cost per transaction trends to zero. Same revenue line, fundamentally different P&L.

Brand

Hudson is the first named AI broker in New York. The first to win consumer trust in the category sets the default — like Stripe did for payments. Editorial, intelligence-driven, not another portal.

Market sizing — what conviction buys
US residential commissions · TAM$100B / yr
0.1% capture$100M GCI
NYC luxury sales · beachhead$8B GTV / yr
50 deals · Year 1$9M GCI
US brokerages · SaaS110,000 firms
1% × $30K ARR$33M ARR
Off-market graph · licensingUncapped
API margin70%+
Competition

Everyone else is fighting the wrong battle.

Every existing competitor is structurally incapable of doing what Signal Homes does — not because they lack the engineers, but because their business models will not let them.

Compass
Better CRM for 30,000 agents. Operating margin: −13%.
won't cannibalize agent splits
Zillow
Advertising marketplace. Sells leads back to agents.
can't operate off-market without breaking ad revenue
StreetEasy
Owned by Zillow. Same structural conflict.
no agents in the buildings — no ground truth
Douglas Elliman
$8.6B GTV. $16.3M net loss. Pure headcount.
no proprietary tech, no data graph, no IP
Redfin
Discount commission. Salaried agents.
race-to-the-bottom pricing — no premium
Side
Back-office for boutique brokerages.
B2B only — no consumer brand, no graph
The founder

A cross-disciplinary operator.
Built for this exact moment.

Matthew Hars

Founder and sole inventor. NYC licensed real estate salesperson with thirteen-plus years closing luxury Manhattan and Brooklyn transactions on a top-producing team.

Earlier career spans network television production (American Idol · The Voice · America's Got Talent), accepted candidacies for the Foreign Service track and the U.S. Naval Academy, and active operation of Hars Investigations — a digital forensics and OSINT practice that directly informs counterparty intelligence in the patent.

Bilingual English and French. Cognitive profile formally documented for exceptional auditory-verbal processing and pattern recognition — strengths that map directly to decoding luxury real estate conversations into structured data.

Domain

Closes the deals the platform is built around. Knows what's actually broken from inside the trade.

Compliance instinct

IC-track and FBI-method training. Builds the regulatory state machine first, not last.

Investigative tradecraft

Operates an active OSINT practice that informs counterparty intelligence directly.

Sole inventor

All IP and equity consolidated. No founder disputes possible.

Where we are

Patent filed. Brand live. Ready to ship.

We are not raising to find product-market fit. We are raising to close the NYC luxury beachhead, file the non-provisional, and instrument the data graph that becomes the moat.

— 01 · Done

IP foundation

Provisional patent filed May 2026. 283 pages. Sixteen subsystems claimed. Priority date locked for the twelve-month non-provisional window.

— 02 · Done

Brand live

signalhomes.ai deployed. Hudson conversational surface live. Editorial brand system established and consistent across all surfaces.

— 03 · Underway

Cohort recruitment

Twelve founding agents from top NYC luxury teams. Equity-bound. Hybrid commission and equity structure designed.

— 04 · Next

Entity · DOS · E&O

Stripe Atlas formation. NY DOS brokerage license. REBNY membership. E&O insurance. Non-provisional patent prosecution.

The math

What conviction returns.

The public market values Compass at $5.3B on negative margins. Anywhere's last private comp was $7.3B. The category is already being priced at ten figures — by the wrong people, with the wrong cost structures.

Your check
At seed
After A + B
At $3B exit
At $5B exit
$50K
0.5%
0.34%
$10.2M
$17M
$100K
1.0%
0.68%
$20.4M
$34M
$250K
2.5%
1.7%
$51M
$85M
$500K
5.0%
3.4%
$102M
$170M
$1M
10.0%
6.8%
$204M
$340M

Illustrative. Assumes $1.5M pre-seed at $10M post, 20% Series A dilution, 15% Series B dilution. Exit comps: Compass ($5.3B public, current); Anywhere Real Estate ($7.3B Compass acquisition); Zillow ($16.8B public). Past results do not guarantee future outcomes. This is a risk asset; investors must be able to lose the entire principal.

The headline
$102M

A $500K check today.
Returned at a Compass-comparable exit. 200× multiple.

We're not asking you to believe Signal Homes will be the biggest brokerage on earth. We're asking whether reaching half of Compass's current public valuation is plausible. We think it's the floor.

The ask

$1.5M on a SAFE.
Eighteen months to revenue and Series A.

We close the NYC luxury beachhead, file the non-provisional patent, ship Hudson v1, and instrument the shadow graph. At month eighteen we raise a Series A at a markup driven by revenue and IP — not hype.

Use of funds
Allocation
Outcome by month 18
NYC operating beachhead
$650K
50 luxury closings · ≈ $9M GCI
Platform engineering
$500K
Hudson v1 · shadow graph · ground-truth ingestion
Patent · non-provisional
$150K
Full utility filing · PCT international option
Compliance · DOS · REBNY · E&O
$100K
Operating brokerage across all jurisdictions
Reserve · contingency
$100K
12-month buffer to Series A

The next decade of residential real estate
will be built by one platform.

We've already filed the patent. We're already in the buildings. We're building it.

Talk to Matthew