The Intelligence Layer

The built world is one of the most heavily measured things on earth. Every parcel has a boundary. Every address has a history of permits, sales, renovations and ownership. Every neighbourhood has demographics, schools, businesses, transit and environmental exposure. The data exists. What doesn't exist — or didn't, until recently — is a way to make all of it speak the same language.

That is the real constraint on real estate. Boundaries live in one system, listings in another, permits in a municipal portal, demographics in a census release, energy in a utility record. Each is captured at a different resolution, in a different format, under a different identifier. On its own, each source is an island. Put a hundred islands next to each other and you still don't have a map.

Homicity was founded in 2015 on a simple conviction: the advantage in real estate would eventually belong not to whoever held the most data, but to whoever could connect it. We spent our first years building consumer property search, pivoted to serving the industry directly in 2020, and have spent every year since turning that conviction into infrastructure. This paper explains how.

The core of our methodology is a single idea: every piece of data about the built world can be anchored to a place, and every place can be anchored to one shared hierarchy. Country, province, city, neighbourhood, street, parcel, coordinate — a continuous chain from the national scale down to a single point on the ground.

We call this the location graph. It is not a database of listings; it is the skeleton that everything else hangs on. When a permit, a school catchment, a business registration or a demographic figure enters our platform, the first thing we do is resolve it to its exact place in that graph. A permit stops being a row in a municipal spreadsheet and becomes a fact attached to a parcel, inside a neighbourhood, inside a city. Once everything shares the same coordinates and the same hierarchy, data that used to be incompatible becomes comparable — and comparable data is where insight begins.

This is why we describe Homicity as the intelligence layer rather than a data vendor. A vendor sells you a file. An intelligence layer gives a coordinate context: what is here, what surrounds it, how it compares, and how it's changing. The graph is what turns a point into an answer.

Every layer in the platform goes through the same discipline, because the value of connected data collapses the moment one source is sloppy. The pipeline is deliberately unglamorous.

We start from authoritative sources — public records, licensed datasets and partner feeds — never client data. We standardize relentlessly: normalizing addresses, reconciling identifiers, and resolving each record to a coordinate and a place in the graph. Where ground truth is incomplete, we use statistical modelling and machine learning to estimate attributes at the neighbourhood level, and we are honest about the difference between a measured value and a modelled one. Then we verify, monitor for drift, and ship it as a layer teams can query through one consistent interface.

The distinction between measured and estimated runs through everything we publish. Rates, boundaries and permit records are stated as fact. Modelled attributes and market trends are labelled as estimates. We would rather be precise than impressive — a principle we hold ourselves to in our research reports as much as in the product.

Location data is powerful precisely because it is intimate — it describes where people live. That is exactly why the responsibility has to be built into the architecture rather than bolted on at the end. Our approach rests on a few non-negotiables.

We build on public, licensed and partner sources, and we share aggregated insight rather than raw personal exposure. Client data is never used to train models that are shared across clients. AI supports expert judgment; it does not replace human accountability for the final decision. And we host in Canadian data centres, aligned with PIPEDA and applicable provincial privacy law.

None of this is a marketing position. It is a design constraint that shapes which datasets we will and won't build, how we aggregate, and how we deploy AI. Data that can't be handled responsibly is data we don't ship.

Most property data answers where and how much. The more interesting questions are what a place is actually like to live in, what it truly costs, and where it's heading. Those answers require data types the industry has historically ignored because they were hard to collect and harder to connect. They are exactly the layers we're investing in.

We think the next decade of real estate intelligence is built on signals like these — each one meaningful on its own, and far more meaningful once connected to the rest of the graph.

A layer in isolation is useful. Layers connected through one graph are transformative. The whole point of resolving everything to the same hierarchy is that a single coordinate can return the full picture at once: the boundary it sits in, the demographics around it, the permits nearby, the schools it feeds, the energy it uses, the risk it carries — through one query, one interface, one source of truth.

That is what Neighbourly.io delivers. For a lender, an address resolves to risk and value in real time. For a brokerage or portal, location context is what turns a listing into a decision. For a proptech team, it's the difference between shipping in weeks and stitching sources together for months. The connective tissue is doing the work that used to take a data team.

It is also why we treat the platform as infrastructure rather than a feature. Infrastructure is judged by what other people can build on it — and the measure of the intelligence layer is not how impressive our own demo looks, but how much faster everyone else can move because it exists.

Real estate runs on information asymmetry. Professionals and institutions have always had data that buyers, sellers and owners never see — while those people make the biggest financial decisions of their lives with a fraction of the picture. We think that's backwards, and the platform that serves the industry is the same one that can level the field.

So we deliberately open parts of it up. Free tools let anyone look up a postal code or neighbourhood's demographics with no account and no cost. Thousands of open location guides put real market and demographic context in front of the public. Verified energy data shows the true cost of a home before someone commits to it. And consumer products like Homeprint and Houzey put property and neighbourhood data directly into the hands of the people making the decision.

Better data leads to better decisions, and better decisions are good for everyone — not just the market. That belief is why Data for Good is a stated part of what we build, not a side project.

We commercialized Neighbourly in 2024, and in 2026 we're building its next generation — deeper layers, richer connections and grounded AI that reasons over verified property data rather than guesswork. The direction is consistent with everything above: connect more of the built world, hold the same standard of honesty about what's measured versus modelled, and keep the responsibility built into the architecture.

If there is one idea to take from this paper, it is that the future of real estate intelligence is not about collecting more data. It is about connecting the data that already exists, responsibly, into something people and machines can actually trust. That is the intelligence layer. It's what we've been building since 2015, and it's what we'll keep building.

This is a living document from Homicity Research. Facts about our platform are stated as fact; forward-looking statements are framed as our direction and belief; anything about prices or markets elsewhere in our research is directional and labelled as an estimate.