.png)
Today EliseAI is announcing a $350M funding round at a $4B valuation, led by Andreessen Horowitz (a16z) and Bessemer Venture Partners, with participation from OTPP, Sapphire Ventures, and Navitas Capital.
Over the past few years, EliseAI has become the predominant AI company in housing. Few other AI companies have embedded themselves as deeply in an essential services industry:
- 6.5M units, 1 in 5 multifamily apartments in the US
- 30M unique renters reached
- $200M+ in ARR
On those units, we’re seeing economic effects that weren’t possible before:
- Occupancy is 2 points higher, effectively adding supply in a housing shortage
- Residents are 7 points more likely to renew
- Net operating income is up as much as 20%
These numbers show, for the first time, that AI can change the economics of a foundational industry at scale.
.svg)
I think they also point toward a bigger idea about the intelligence age: To ensure the benefits of AI reach everyone, it isn’t enough to just make a better model, like OpenAI and Anthropic are doing, or even put a “superintelligence in everyone’s pocket”, the idea behind Muse.
To ensure the benefits of AI reach everyone, we have to work incredibly hard to bring AI to the essential goods and services - like housing, healthcare, food, and education - that determine most people’s quality of life and the cost of living.
These industries are the most complex, the most bespoke, the most entrenched, and the hardest - but they’re also the most important. And they won’t absorb this technology without a tremendous act of building.
That is our mission at Elise. Make sure this extraordinary technology raises the standard of ordinary life for everyone, including people who never open up a chatbot. And do this by solving the hardest applied AI problems there are in the parts of the economy that lift all boats.
Your Life Depends on More Than How Smart You Are
A lot of serious people are working on the same general problem. How do we make sure the most powerful technology of all time benefits everyone, and not just a chosen few?
Step One: The frontier labs are doing their part by racing to build an aligned superintelligence that manages the transition gracefully.
Step Two: Zuck and the whole open-source ecosystem are trying to ensure that everyone has access to the best intelligence. This is the idea behind Muse and The Future is for Everyone: a personal superintelligence in everyone’s pocket.
At EliseAI, we believe there is a third necessary step. It’s critical that everyone has access to strong AI, but it doesn’t solve everything. Because the quality of your life depends on more than how smart you are.
Consider what happens if you’re looking for an apartment. You have a really smart AI in your pocket. This assistant can go above and beyond to help you find an apartment. It can look at every listing in San Francisco, find the ones you’ll like most, and even negotiate the lease.
But there are also things it can’t do. It can’t make more apartments available to you or change their rents. It can’t change the quality of a building once you live in it. And it can’t magically make more supply appear out of thin air.
In other words, the quality of your life depends on more than how smart you are. It depends on the world of systems and resources around you.
This reliance on the world around you gets worse because of Baumol’s cost disease. For fifty years, technological progress has been lopsided. Computing made televisions, phones, and software cheaper.
Housing, healthcare, childcare, and education got more expensive every year. This phenomenon, where prices for discretionary goods go down while prices for essential goods go up, happens when productivity surges in one part of the economy, wages rise everywhere to keep up, and the sectors where productivity didn’t improve get relatively more expensive forever.
Practically, what this means is that skyrocketing productivity in Silicon Valley may make people a lot of money, but the cost of living might rise right in tandem - for everyone. So your superintelligent assistant would come back to you with some… really expensive apartment options.
Of course, AI has the potential to reverse Baumol’s cost disease, unlike every technology before it. The reason is that AI isn’t a good, it’s a service: it does the work. So it can make providing essential services like housing and healthcare more efficient and bring down prices as a result.
But to do that, AI has to diffuse - not just theoretically but actually - into the essential industries that determine the cost of living.
There’s just one problem: these are the hardest industries in the world to diffuse a technology into.
The Hardest Industries For Diffusion
As Dario has long said, AI diffusion will be “fast but not instantaneous”, even once takeoff starts. Diffusion will take work. And it will be slowest in the industries that are most inertial - the most physical, the most bespoke, the most resistant to change.
There are a few reasons that essential services industries fit the bill for “hard diffusion” alarmingly well. First, demand is inelastic (everyone needs a home), so there’s less pressure to improve. Second, service is local (a building is a physical thing), so it’s hard to scale with technology and realize efficiency gains. Third, mistakes are costly, and the bar for reliability is higher (and of course, generative AI is probabilistic).
But the “hard diffusion” industries also hold the biggest prize. Because they’re so hard, these industries have been left behind by previous technologies, and so they have the most headroom to move prices. There are enormous returns in using AI to change how these industries operate in fundamental ways that also improve everyone’s quality of life.
Let’s look at what we’re doing in housing as an example.
The Operating Costs Are Too Damn High
Housing is the largest expense for most households. We clearly need to build more homes, particularly in places with severe shortages like San Francisco and New York. Most housing crisis discourse is focused on this supply problem, including zoning laws and NIMBYism.
But there’s another equally important problem people aren’t talking about: Today, operating a building is a really hard business. Margins are tiny, and this both impacts supply and directly drives up prices.
As Jay Parsons has written, a third of US households earn too little to afford the average operating cost of a US apartment, much less its rent. This is because a typical operating cost for a US apartment is $600+ per month, driven up by highly manual and non-scalable processes that largely haven’t changed since on-prem software came to the industry in the early 1980s.
EliseAI, as mentioned above, is helping change this at scale, to the tune of 20% higher net operating income in some cases, and a stack of AI products that impact 58% of a typical operator’s P&L.
Unafraid of the Real World
Because housing is hard, this has been painstaking work with a lot of challenging problems baked in. It goes beyond ideas like “industry context” and “in-depth workflows” that are overused in the apps-layer conversation.
You might think you can combine an LLM with a system of record and some context on how to sell apartments and get a pretty good Leasing AI product. But this would never be enough to serve enterprise customers or change how an entire industry does business.
Here’s a tiny sampling of what we’ve had to build to do just that:
- Self-learning - when there is no database of all the relevant information, the AI has to teach itself
- In-house models - for tasks that the frontier models can’t do, we train in-house models (for example, renewal likelihood prediction and address matching)
- Physical world autonomy - truly AI-guided tours require smart lock integrations and wayfinding across an entire building
- Multiple interconnected model types - reducing voice latency without impacting response quality means mixing smaller open-source models with larger reasoning models and generating composite responses
- Regulatory compliance - affordable housing operations require 200-page certification documents that AI needs to generate, help renters fill out, and approve with zero margin for error
- Reinventing the operating model - automation in itself isn’t enough; it has to change how a company is structured and does business. We reinvented the industry’s CRM and built products specifically for the new, more efficient AI-based operating models our customers are creating.
From the outside, it looks like a chatbot that texts renters about apartments. Underneath, it's some of the hardest applied AI work there is. That’s what’s attracted PhDs, engineers from frontier labs, and others who want to work on the problems that stand between AI and the real world.
That’s our purpose at Elise: diffusing AI to the most important parts of the economy, starting with housing, by solving the hardest applied AI problems that there are. So that abundance reaches everywhere, including the most essential goods and services. And so that this extraordinary technology raises the standard of ordinary life for everyone.
This round gives us the ability to move faster. The people who join us now will decide how far we can go. If you want to take on difficult work, learn an industry deeply, and build things that change what happens outside the digital world - there’s a lot left to build.
LFB.