HouseCanary

Who they are now

A mortgage gets approved at 9am. A canary saw it coming.

Somewhere right now, a lender is deciding what a house is worth. Not by sending a person to walk the property with a clipboard - that takes days - but by sending a query to a server farm. Milliseconds later, an answer comes back: a number, a confidence band, a forecast. The house has never been visited. The estimate is usually within a few percent of what it will actually sell for. The system behind that answer, more often than the public realizes, is HouseCanary.

HouseCanary is an AI-powered real estate analytics company and a licensed national brokerage, headquartered in San Francisco and run as a fully remote operation. It sells automated valuation models, home-price forecasts, and property-data APIs to the people who move money through housing - lenders, capital-markets desks, single-family-rental investors - and it runs a consumer-facing platform, ComeHome, that those same institutions wrap their brand around. It is, in plain terms, a data company that happens to be pointed at the biggest asset class in the country.

The problem they saw

Housing is enormous. The data was a mess.

Here is the uncomfortable thing about residential real estate: it is worth trillions, it touches nearly every household, and for most of its history it has been documented like a yard sale. County records on paper. Inconsistent formats. Gaps where data should be. No common standard between one jurisdiction and the next. For an industry that decides how much people can borrow against the roof over their heads, that is a remarkable amount of guesswork.

The cost of that mess is not abstract. Slow appraisals delay closings. Inconsistent valuations introduce risk into mortgage portfolios. Investors buying thousands of rental homes need to price them faster than any human appraisal pipeline can manage. The market was big enough to matter and disorganized enough to break - which, depending on your temperament, is either a crisis or an opportunity.

The founders' bet

A consultant and an economist walk into a collapsing market.

The idea traces to 2008. Jeremy Sicklick was a partner and managing director at Boston Consulting Group, helping large real estate investors deploy billions of dollars. Trying to value those portfolios as the market cratered, he kept hitting the same wall: the underlying data was incomplete, unstandardized, and slow. You cannot manage what you cannot measure, and housing - the thing American wealth is mostly stored in - was barely measured at all.

Sicklick co-founded HouseCanary in 2013 with Christopher Stroud, an economist whose doctoral research focused on dynamic models of financial risk. Stroud, then in his late twenties, took on the research engine: building predictive models that could learn the tangled relationships between location, condition, market momentum, and price. The bet was simple to state and hard to execute - digitize the country's property data, then teach machines to value it better than the status quo.

It was not a cheap bet, and the people who funded it were not ordinary. Over five rounds HouseCanary raised roughly $130 million. The cap table reads like a dinner party no one would believe: Hillspire, the family office of Alphabet's Eric Schmidt; PSP Capital, founded by former Commerce Secretary Penny Pritzker; and Bryant Stibel, co-founded by Kobe Bryant. The $65 million Series C in February 2020 - led by Morpheus Ventures, Alpha Edison, and PSP Growth - matched everything the company had raised before it, combined.

The product

Four letters that run the machine: A-V-M.

At the core sits the automated valuation model. An AVM is a machine-learning system that estimates a home's value from data alone - comparable sales, neighborhood signals, property characteristics, market momentum - without a human walking through the door. HouseCanary's covers roughly 136 million properties and reports a median absolute error of about 2.7%. Its twelve-month home-price forecast reports a median error near 1.7%. Those are not marketing numbers you can wave away; they are the difference between a loan that prices correctly and one that does not.

Around the AVM, the company built the things institutions actually buy. Data Explorer and a set of APIs let lenders and developers pull property intelligence straight into their own software. Value reports and comparables support individual lending decisions. Market forecasts and home-price indices feed portfolio monitoring and risk teams. And then there is ComeHome - the part consumers might actually touch.

ComeHome is the consumer-facing layer, and it is clever precisely because it hides the machinery. A lender or brokerage embeds a co-branded home-search and homeowner dashboard. Buyers browse listings; existing homeowners watch their property's estimated value drift up and down. The institution gets engagement and retention; HouseCanary gets distribution. Everybody behaves as if they are using a friendly app, while underneath, the same valuation models that price mortgage portfolios are quietly doing the math.

The proof

Accuracy you can put in a chart.

Skeptics are right to ask: compared to what? A valuation model is only useful if its errors are small and honest. The argument HouseCanary makes is numerical, so here it is, drawn to scale.

The proof is also in the company it keeps. Fitch Ratings recognized HouseCanary as a leading provider of real estate valuations in 2020 - which matters, because ratings agencies are paid to be hard to impress. Ally Financial put ComeHome on its own website. And in June 2026, after a quiet 2025 pilot across eight markets, Google chose HouseCanary to power the national expansion of its home-listing discovery program, surfacing MLS broker listings directly inside Google mobile search. When the largest search engine on earth wants housing data flowing through its results, the vendor it picks tells you something.

The mission

Make the biggest market legible.

Strip away the product names and the mission is almost stubbornly plain: modernize residential real estate so that valuation, lending, and investment decisions run on accurate data instead of educated guesses. That is not a slogan that wins design awards. It is, however, the kind of problem that compounds - every additional property indexed, every model improvement, makes the next decision a little less of a gamble.

There is a reason the logo is a canary perched on a roof. The canary in the coal mine was an early-warning system - a small creature whose distress signaled danger humans could not yet see. Point that idea at housing and you get the company's quiet ambition: to see risk in the market before it becomes everyone's problem. Whether a small yellow bird can carry the weight of a $2 trillion asset class is a fair question. So far it has not fallen off the roof.

Why it matters tomorrow

Back to that 9am mortgage.

Return to where we started. A lender approves a mortgage in the morning, and the valuation underneath it never required a person to drive anywhere. A decade ago that was a slow, manual, error-prone ritual. Now it is a query. The clipboard did not disappear because someone declared it obsolete - it disappeared because the data finally got good enough to trust, and a company in San Francisco spent more than a decade making it so.

HouseCanary does not own the housing market, and it is not the only firm racing to model it - CoreLogic, ICE/Black Knight, Zillow and others are all in the chase. But the direction is set. The largest asset class in America is becoming legible, queryable, forecastable. The canary is still on the roof, still watching. And somewhere, right now, another house is being valued without anyone knocking on the door.