Thoughts on "Opendoor’s AI Journey"
The Good: Huge Sea Change in Tech-Orientation and Messaging
The first and most obvious take-away is that Opendoor is fully embracing both AI and retail as a strategic business and marketing lever (better late than never!)
On the tech side, the shift has gone from talking about adding "school quality" and "active comps" (duh?) in the Q1 2025 Earnings Call to now having a system that can identify how a specific tree on the property might impact valuation. This feels like 10x improvement, at least, in the "tech-sophistication" of the company.
I also love the specificity that Shrisha is able to highlight (cabinets, paint, odors, etc.). This shows that he is detail-oriented and in the weeds - a hugely critical factor if Opendoor is to move fast (think Elon, Jensen, Zuck) as it allows him to quickly make "the final call" on important unblocking technical and strategic decisions.
It is also clear that with a new leader at the helm and the excitement of retail sentiment, this has catalyzed a new mode of operations and experimentation within the group. In fact, even the meme accompanying the article gives us some signal around how the company is reorienting - not just embracing AI, but also humor and retail feedback. Drizzy partnership soon?
The Bad: So What?
The biggest thing I felt was lacking in the article was the "so what" ie. How is AI going to actually impact business metrics.
Here's what I would love to see shared next:
1) How is AI going to power buybox expansion?
If RiskAI can truly do something as complex and nuanced as identify how a specific tree can impact a property's valuation, there's absolutely no reason why Opendoor can't expand its buybox to more expensive and / or "bespoke" homes.
The whole purpose of the pricing technology is that it can capture each home's unique characteristics and allow Opendoor to offer confidently on all the homes in a market. From first principles, if you have a technology that can identify the "unique feel" and characteristics of a home (maybe supported by an human agent walkthrough) - then a $1M home should really be no harder than a $400K one.
I would love to see a "conquest map" of how Opendoor is going to use AI enhancements to penetrate deeper into existing markets (move up price points) and also tackle new ones (Boston, New York, Bay Area) - this is what will really move the needle on the business.
2) How is AI going to power pricing accuracy and reduce spreads / increase conversion?
Let's pretend we had a perfect crystal ball (a "palantir" if you will 😉) that would allow us, for every single home, know exactly WHEN and HOW MUCH it would sell for (if we bought the home today). What would this unlock?
At the right scale and cost-structure, having 100% accurate expected margins on a home would allow Opendoor to set a spread that would almost certainly be better than the traditional listing fee (<6%) - simply because spread is a proxy for risk / uncertainty.
Lower (or similar) Fees + Fair Cash Offer + Certainty + On Your Own Timeline - I believe that's what Keith would call a "no-brainer" customer value proposition and something that would allow Opendoor to truly unlock the full TAM of the industry (not 5% conversion but 50%)
I think this should be the stated ultimate goal that Opendoor pursues with AI. To 100x reduce the uncertainty around its pricing models so that it can build towards that "all knowing crystal ball."
Unfortunately, however, this wasn't the clarity I got from the article. The "business impact" language in the article was all relatively generic:
"This improves offers, makes the process more seamless, and builds trust"
"Will let us provide clear guidance to sellers and agents alike, and ultimately stronger offers that fully recognize the uniqueness of an individual home.
"Our work underpins our ability to reach more customers with innovative products like Cash Plus and empower our Key Agent network to provide an unparalleled service."
Understood that this is potentially only the first article in a series but would love to hear more about the financial and operational impact the team is driving with AI (what % buybox expansion, what increase in offer accuracy)
Maybe at an investor day? 🌽
Final Thoughts: Execution and Speed is the Moat; Not the Data.
Shrisha touted that the team built RiskAI and Repair Co-Pilot in days / weeks - is this really a good thing though?
I have a sneaking suspicion that this is actually a sign that Opendoor has potentially less of a data / tech moat than previously thought and that modern LLM's are just that damn good.
The article mentions Opendoor is currently at 250K closed transactions and 2M first offers. Back in 2022, they were probably already at ~200K and 1.5M. Is there some signal from the most recent 50K transactions that the first 200K were missing? Seems unlikely. More likely is that the pricing algorithms, as currently designed, have simply hit a wall in terms of how much better they can get.
Enter modern AI models (language, image, video). Having tested them, I believe that leading-edge AI's (with maybe a tiny bit of context and tuning) are probably already pretty close to Opendoor's internal models. If you think about it, they've already been "pre-trained" on millions if not billions of data points on how to think about home valuation, neighborhood characteristics, and "pricing" heuristics.
As a result, could an AI-native start-up replicate Opendoor's pricing algorithm in a few weeks? Potentially.
That's why, the only real moat now is execution, iteration speed and cognitive flexibility - a willingness to try new things and discard the old. Even now, new competitors are starting to pop up (Roam anyone?) and Opendoor won't be the only (real) player in the space for long.
"I think the only moat in our category is speed. It's learning where the dead bodies are, learning how to compound" - Windsurf CEO Varun on 20VC
As a parting thought, I want to acknowledge that Eric's involvement in the company, Carrie moving on, and the retail fervor has catalyzed a truly "once-in-a-generation" opportunity for the company and investors. I feel extremely fortunate to be able to participate in something so unique and share my thoughts with the $OPEN army.
Even at $5 though, we're still only at the pre-game...
Party hasn't even started yet 🚀
