Jeff Dvorett is President of Midwood Investment & Development, a vertically integrated owner, developer, and manager of mixed-use real estate founded in 1925. Midwood owns and operates more than 140 assets — predominantly residential and retail — and has several million square feet in its development pipeline. Headquartered in New York City with operations in the Northeast Corridor, Pittsburgh, and Los Angeles, the firm has spent the last several years building a proprietary data analytics platform that helps it source deals proactively rather than react to inbound flow.
Jeff will be on stage at Blueprint September 22–24 in Las Vegas, joining a panel hosted by Rebuild Capital founder Jeremy Kaner on the changing landscape of capital and innovation in proptech and construction tech.
We caught up with him ahead of the conference to talk about how Midwood evaluates technology, what’s actually clearing the ROI bar, the foundation-model threat to point solutions, and where the next 100 years of a 100-year-old firm goes from here.
To kick us off, tell us a bit about Midwood and your role there.
Midwood was founded in 1925. We’re a vertically integrated owner, developer, and operator of mixed use real estate with over 140 assets, predominantly in residential and retail asset classes, and several million square feet in our development pipeline. We’re based in New York City but have offices and operations centralized in the Northeast, plus Pittsburgh and Los Angeles.
I started here as head of development and over time stepped into the role of President. I have one foot in development and one foot in our ongoing operations.
For a firm celebrating its 101st year, we have a diverse portfolio. But as we look forward, we’re focused on vertical mixed-use multifamily rental above retail and parking. We are also focused on retail from our portfolio of well-located neighborhood street retail, like our assets on Walnut and Chestnut Streets in Philadelphia to high performing shopping centers. We developed the Shops at Sportsmen’s Lodge in Studio City — a 100,000-square-foot beautifully landscaped neighborhood retail center with an incredible tenancy, anchored by one of the top Equinoxes in the country, Erewhon, and three Sugarfish-group restaurants along the Los Angeles River.
Being nimble in retail has been important. There’s a lot of opportunities to develop or acquire new multifamily, but new retail opportunities are more diverse. The opportunity in Studio City was very unique.
Midwood is celebrating its 100th year, but the proprietary data analytics platform you’ve built is very much a 2020s story. What changed?
Being data-forward has been a core value at Midwood since inception. In the 10 years I’ve been here, our CEO John Usdan and our Chief Investment Officer, Teodora Zobel, have really taken it to the next level.
What we realized — particularly during the pandemic — is that data doesn’t just inform decisions. It helps us speed up and streamline the decision-making process itself. Time is so often critical in real estate, so having a lens through which to analyze patterns gives us a competitive advantage.
The tools we’ve developed also allow us to proactively source opportunities instead of just reacting to inbound deals. If we’re looking for a site of a certain size in New York with zoning that allows for a particular type of use — commercial, retail, or residential — we can narrow the data set quickly. That’s the top-level work. From there we get into census information, demographic trends, capital inflows and outflows, sometimes down to the block level.
We can say: this is a neighborhood we had a hypothesis about — does the data support it? Or, alternatively: we never would have thought wealth generation was happening in this zip code — let’s dig in and get some boots on the ground. Sometimes that turns up something for sale that’s interesting given the backdrop. There is no substitute for in person visual observation, but this really helps us narrow down where to look.
You operate across investment, development, and management. When you evaluate a new piece of proptech or construction tech, who actually has to say “yes” — and how has that changed in the last 24 months?
One of our core principles is investing in our team. The people who live in the business every day know it best and are best suited to advocate for the innovation they’re going to use. Final approval sits with me and the leadership team, but it warms my heart when a development associate, a property manager, or an accountant says, “This is an area we’ve struggled with — it would really help if we could find a way to do it better.”
Decision timelines depend on a bunch of factors. Is there high risk and high opportunity cost in implementation? Or can we run a pilot on the side, and if it doesn’t work, no harm, no foul?
Northspyre is a good example. We labored over that decision for a long time. We knew Excel wasn’t the right tool for development project management, but the perceived opportunity cost of pivoting away from something everybody was comfortable with felt enormous. The push to actually do it came from our development project managers — they wanted the best tool at their disposal and were willing to invest the time in learning how to use it. That was the unlock.
Jeremy Kaner’s thesis with Rebuild is that there’s a capital gap between $50M and $500M for tech outcomes that don’t fit the venture model anymore. From the operator side of the table, do you feel that gap?
I believed in what Jeremy was saying so much that I decided to invest in his business. So, I’m a believer.
The math has changed. As owners, we’re always looking to expand margins, gain efficiency, and improve operations. The idea that we’d have access to technology with no deployment risk, existing customer relationships, and a clear, already-proven ROI — and that the underlying companies are more durable operating businesses — is genuinely compelling. There’s less risk on that company not being there at some point in the future.
That’s something we’re always concerned about: is the company we’re investing our opportunity costs into going to be around six to twelve months from now, after we’ve started to depend on them? Durable operating companies — the kind Rebuild is targeting — mitigate that risk in a way that pure venture-stage proptech can’t.
The market has clearly shifted. Proptech now has to deliver core ROI that impacts cash flow, not just future-looking optionality. Give us a concrete example of a tool or category that cleared that bar at Midwood, and one that didn’t.
One that’s clearly cleared the bar — and we’re at the stage of figuring out which partner to work with — is sales-enhancement tools in multifamily leasing. Grotto, Rilla, and others in that space.
We’ve come out of the world where you secret-shop your multifamily building twice a year, tour, talk to your sales people, and see how they’re doing. The value proposition Grotto and Rilla have proven, both in our industry and in other sales organizations, is really compelling. The measurement is also clean: we can roll this out and, three months later, determine whether there was a measurable ROI on the deployment.
Other things we get excited about are much harder to evaluate. As soon as you introduce more variables into the technology you’re bringing to bear — say, a tool that claims to improve resident retention — how do you control for just that variable for somebody who’s lived in a unit for a year? That’s where I struggle, and where I think a lot of the industry struggles.
There’s a growing argument that ChatGPT and a handful of horizontal AI tools are replacing the bottom 30% of the proptech stack. Are you seeing that play out at Midwood — and which categories of vendors are most at risk of being collapsed into general-purpose AI?
At some level it has to happen. For the lower-hanging-fruit problems we solve with conventional software, vibecoding or using GPT or its competitors to replicate that functionality is increasingly feasible. The more general-purpose the software, the easier that disruption becomes.
Frankly, I haven’t seen it happen yet to most of our tech stack. But I believe it’s coming. It has to. I don’t think it’ll be a complete disruption — there’s a level of data integration, and some of these technologies go beyond pure software platforms in ways that will be much harder to replicate and take out.
The other side of the argument: sometimes the existing solution is good enough. It’s tried and true. It’s worth using until something much better comes along. But 30% sounds about right in terms of what’s going to happen to the more generalized software platforms.
Midwood targets value-add, redevelopment, and ground-up across the Northeast Corridor and California. Construction tech has been notoriously hard to scale. What’s actually working on your job sites right now versus what’s just demo-ware?
OpenSpace.ai is terrific. I haven’t used it long enough to get what I believe the full value proposition is going to be — which goes well beyond construction documentation and site supervision — but having that formal visual record is enormously valuable.
I can’t tell you how many times I’ve wished I had a picture of what something looked like behind a wall after I built it. To have that as a record set, with all the other things they’re building to layer on top of it, is a very clearly defined value for us.
Your site selection model leans heavily on hyper-local data analytics to find undervalued neighborhoods before the market does. How do you think about data as a moat — is proprietary data still defensible, or has the AI era made everyone’s data more interchangeable?
There’s always been data available to some degree. The example I think about is our CEO telling me that, starting in the early 80s, he was getting publicly available data on subway ridership and foot traffic at urban intersections. The data was out there. What mattered was knowing what to look for and being able to understand the patterns.
If you didn’t have a methodology to go through the data and find those patterns, it wasn’t very useful. Even today, with more data available than ever, the methodology is as much of a differentiator as the dataset itself. Pattern recognition — knowing what signals to take seriously and how to sort through them — is more proprietary than the raw data.
Will an algorithm eventually be able to do that too? I wouldn’t want to speculate on the limits of what AI will eventually be able to do. Computational skills are increasing at such a rate that it’s probably inevitable that pattern-recognition layer can be created. I think there is a lot of runway before we get there so I am more focused on the medium term.
A lot of proptech founders pitch operators every week. What’s the one thing the best founders understand about how owner-operators actually think — and what’s the most common mistake the rest make?
Founders have this incredible sense of energy and optimism that makes meeting with them super energizing. I love meeting with founders.
The pitfall is that sometimes they get so excited about the technology they’re pitching that it leads to tangents — and solutions to problems that aren’t really problems, or aren’t the problems on my mind. It becomes, “This technology is really cool, look at what we can use it to do,” and it gets away from what problem are you actually solving.
Being very clear on the value proposition the technology brings to a customer, and being able to articulate it crisply, is first and foremost for a successful pitch. When founders get too excited about the tech and the pitch goes off course, that’s where it breaks down.
Recaps, mergers, hybrid venture-PE structures — your panel is going to dig into all of it. As a long-term operator with relationships across the capital stack, where do you think proptech consolidation is heading over the next 24 months, and what does that mean for the operators who depend on these tools?
It’s certainly happening. And as a user, people get fatigued. There are a lot of point solutions in the portfolio.
If it’s a vertical rollup that increases functionality and gives you more of a one-stop-shopping experience, that’s compelling from a user standpoint. Where I get more cautious is horizontal consolidation — competitors buying competitors and consolidating overlapping functionality. That goes back to the question of what it means for us as users: will the functionality we rely on remain?
It could go a lot of directions. The positive spin: I was a customer of XYZ, the acquirer adds capability, and I benefit from the expanded functionality. The negative: the company I was relying on gets acquired, and they stop providing the level of service or the specific functionality I depended on.
As users, we’re always thinking about risk management with anything in the tech stack. But for the most part, I’m optimistic these consolidations create opportunities for us to get better outcomes with the tech we’re using. It’s better than the alternative of these companies going to zero and the tool just not working one day.
Looking ahead to your next 100 years — and more practically, the next 12 months — what’s the one capability or category you’re most actively trying to build, buy, or partner your way into right now?
Because we’ve been around for so long, we have a tremendous amount of historical data and archives. We’re in the middle of a corporate-wide effort to better organize and streamline that information — adding metadata, adding tagging, structuring things in ways I didn’t even know were possible a few months ago.
The intent is twofold. First, we want to have fast and accurate recall of our files and have the best possible records. Second — and this is where it gets exciting — as we move through that process, we can build or use tools that layer on top of those organized records. As we make sure all of our property information is in the same format, in the same SharePoint setup, what value can we get from layering AI tools on top of that to audit the things we’re doing and create new, better workflows?
There’s a tremendous amount of organizational, audit and compliance work we’re really focused on as we scale, and I think AI can take us to the next level on it. That’s where we’re spending the most energy right now.
