Nick Deveau, Co-Founder and CEO of Grotto AI – Interview Series

Nick Deveau is Co-Founder and CEO of Grotto AI, an AI platform that helps multifamily operators reduce vacancy loss by supercharging their leasing teams through real-time coaching. With 10+ years of experience building AI, including graduate training from Stanford, Nick has driven the development of vertical AI SaaS products for Fortune 500 clients across industries like insurance, healthcare and hiring. Most recently, he co-led the development of the key technology at EvolutionIQ, which drove the company’s $730M acquisition in late 2024, one of the first major vertical AI SaaS exits.

Grotto AI is an AI-powered sales coaching platform built specifically for the multifamily housing industry, helping leasing teams convert more prospects into residents through real-time call guidance, personalized tour coaching, and portfolio-level performance insights. The platform learns from successful leasing interactions and provides agents with guidance on prospect needs, objection handling, unit recommendations, and follow-up actions, while automatically capturing renter information and updating connected customer relationship management and property management systems. Founded by CEO Nick Deveau and CTO Ben Epstein, Grotto aims to improve leasing performance, shorten employee training periods, reduce administrative work, and give property operators greater visibility into renter preferences and conversion drivers.

You spent years building AI systems in healthcare, insurance, hiring, and academic research before co-founding Grotto AI. What convinced you that multifamily leasing was the problem worth dedicating your next company to, and how did the founding vision for Grotto take shape?

I’m a New Yorker, so I think here you either love or hate residential real estate. I happen to love it. And previously, I’d spent my career in data and machine learning, so I went out and asked: what are the huge problems in residential real estate that data can actually solve?

So I got connected to people in the industry and just started asking, “What are the biggest problems you have?” And vacancy loss came up over and over, in every single conversation. It’s simple — it’s the money that apartment owners lose when units sit vacant. And it turns out to be massive: about $30 billion a year in lost rent revenue, and over half a trillion dollars of asset value erased. It’s a huge, huge problem.

Here’s the thing I’ve learned from building companies over the years: some problems are better to build a business around than others. The best criteria is a problem that shows up on your customer’s balance sheet really, really clearly. There are lots of problems worth solving that don’t show up on a balance sheet — and those are much harder to build a great business around, because you can’t easily justify your impact. Vacancy loss is the opposite. It’s an undeniable business case. It stares the customer right in the face. So we said, that’s it. Let’s go solve it.

Before launching Grotto, your team analyzed millions of leasing conversations and found that human behaviors such as curiosity, warmth, and even laughter were stronger predictors of conversion than many traditional sales tactics. What surprised you most in that research?

Honestly, it’s almost what didn’t surprise me. Laughter, rapport building — anyone who’s been in sales, or who’s ever had to convince anyone of anything, knows that relationship building is foundational to getting the job done.

What was surprising was that we backed it up with data at massive scale — and that the data cut against the grain of today’s climate, where the narrative has been “automate everything.” Since we first put these numbers out, we’ve actually watched the narrative shift. It’s moving from “AI should be doing everything” to “AI is great at the grunt work, and because AI does that work, it frees humans up to do the human things that actually move the needle.”

That reframes everything. If relationship building is the priority, then you build tools that supercharge people during those moments, and you hire natural relationship builders and use tech to augment them on the leasing-specific parts of the job.

Many AI startups focus on replacing human work with automation. Grotto has taken the opposite approach, using AI to help leasing agents perform better during critical conversations. Why do you believe augmentation is a more effective strategy than replacement in this market?

It’s simple. The most valuable things that happen during leasing interactions — rapport building, relationship building, even politeness (we have data on that too) — are the things humans still do best. So you need humans doing this stuff.

Once you accept that, the augmentation-versus-replacement debate answers itself. If humans are required for the moments the data says matter most, you have to augment. You can’t just replace.

Grotto’s Guidance platform delivers real-time coaching during calls, tours, and follow-ups. What are the biggest technical challenges involved in providing actionable guidance in the middle of a live human interaction?

Frankly, the hardest challenge stems from delivering a really high-quality user experience — specifically, a low-latency experience that keeps people in the flow of the conversation. In a lot of agentic systems, like chat, you can tolerate a lot of latency. On a live call, you can’t.

It sounds easy in a demo: just put some prompts up telling people what to do. But making sure those prompts are high quality — that they guide people without pulling them out of the flow — is really non-trivial. And ultimately, our impact is mediated by humans. If humans don’t want to use the product, the product doesn’t get used, and we don’t drive impact. So we have to build something people genuinely want to use.

When we first deployed our Call Guidance product, the immediate feedback was: “Guys, this is cool, but these prompts are too long, too slow, they take me out of the flow of my call.” So we put all of that under the microscope and solved for it.

The other piece is setting expectations. Becoming A-plus at leasing calls when you’re starting at a B-minus is a little like rebuilding your basketball shot. It’s going to feel awkward at first. But after a few days of practice and coaching, you come out the other side 10x better than before.

Your career has included building AI systems in highly regulated industries such as healthcare and insurance. How has that experience influenced the way you think about trust, explainability, and responsible AI deployment at Grotto?

Enormously. I’ve spent a decade building B2B vertical AI across hiring, healthcare, and insurance, and the biggest thing that experience gave us is a deep appreciation for change management.

AI products have probabilistic outputs. They’re mostly right, but not always — and when they’re wrong, users can lose trust, even when the product is still way better on average than whatever came before. Getting this messaging right is critical.

And it matters more today than ever. When you deploy into a company, sure, there’s buy-in at the management level. But the end users aren’t necessarily people reading TechCrunch every day. These are often young, fresh-out-of-college leasing agents all over the country who, frankly, have better things to do than track what’s happening in AI. So we have to message directly to them: AI is here to help you — to make you better at your job, make you more money, get you promoted — not to replace you. We literally cannot do this without you.

The other lesson is going shoulder to shoulder with our partners. Most of our customers are real estate companies, not professional tech-deployment companies. So we play a consultative role and directly help roll the tech out — we have a whole forward-deployed motion built around it. A minority of customers prefer something lower touch, which we offer as well, but generally there’s an appreciation for the change-management aspect of this and our customers are eager for our help on it.

Leasing agents often have vastly different styles and personalities. How does Grotto distinguish between coaching best practices and forcing agents into a one-size-fits-all script?

We spend a lot of time on this. We hear the word “flow” a lot — people want to stay in their flow. So the way we think about it is less like giving agents a script they have to follow, and more like giving them a playbook with guideposts: things they need to accomplish at some point in the conversation.

Those guideposts can be hit in many different ways. Take a common one: does the agent ask the prospect an open-ended question? There are endless ways to do that. “Tell me a bit more about that.” “What are some things you can’t live without?” As long as our product understands the semantics of all of those, we can make sure people hit the milestones without dictating the exact words they say.

The multifamily industry has invested heavily in chatbots, self-service tools, and automation. Where do you think the industry has overestimated AI’s capabilities, and where is it still underestimating AI’s potential?

On overestimating: there’ve been a few firms trying to run essentially humanless properties — no human leasing agents, no permanent on-site staff. That vision sounded great in the boardroom. Office and on-site payroll is a huge line item — just cut it, plug in AI. It hasn’t panned out, partly from a feasibility perspective, but also because it’s just not what most of the industry wants. Talk to leaders in this space and they’ll tell you: this is a people business. It’s about relationships. It’s about people finding home and making home. That’s a very clear headwind to total AI takeover of on-site operations.

On underestimating: you can measure underappreciation by lack of investment, and there’s been a total underinvestment in using AI to up-level human team members during critical human-to-human interactions. The most advanced thing we typically see is call scoring: dump a transcript into a rubric and grade it. That’s such an underpowered use of the technology compared to what we’ve found possible.

What’s actually exciting is taking a bottoms-up approach: figuring out what behaviors your top performers deploy that make them great, building new playbooks and rubrics from that, and then guiding people in real time to deploy those behaviors. Most firms are still secret-shopping their properties against a static rubric. We’ve never met anyone doing that type of bottoms-up work themselves — and honestly, it’s quite hard.

There’s one more piece: these human moments should be treated as the central nodes in a prospect journey, with really intelligent automation built around them using the data those moments generate. The market overcorrected toward full autonomy for a while. The right answer is human moments at the center and automation extending out from those nodes where relevant.

Grotto is tackling vacancy loss, a problem that impacts property owners and operators at massive scale. What metrics do your customers care about most, and how do you demonstrate a direct connection between AI guidance and business outcomes?

Vacancy loss is our North Star metric, in large part because it shows up on our customers’ balance sheets. We work with our customers to track a number of leading indicators that roll up into Vacancy loss until we have enough data to measure that North Star. One example of a leading indicator is various conversion rates in the leasing funnel – if these increase, over time vacancy loss should decrease.

We also track certain operational metrics, like time-to-ramp for new hires. With employee turnover being about once every 12-14 months in this industry, the cost of a 3-month ramp period vs 2 week ramp period is enormous (i.e, nearly one-quarter of all on-site payroll could be seen as “not fully productive”), and reducing it should give peace of mind to whoever is footing the bill.

As large language models continue to improve, what new capabilities do you expect AI coaching systems to develop over the next three to five years that are not possible today?

I don’t want to give too much away about our roadmap, but as models get better, faster, smaller, and cheaper, it becomes easier than ever to do really powerful real-time work.

On the coaching side, there’s more and more real-time feedback. Just like you have turn-by-turn directions when you’re driving somewhere, I want really smart turn-by-turn directions for every kind of interaction.

And as always, the UX is the thing to get right. Done wrong, this stuff is distracting and ugly — there are some pretty clear failure modes. But you can get it right. I think about my own sales meetings: sometimes it’s Tuesday at 4pm, I haven’t had my coffee, and there’s just a lot to remember. Having all of that in my field of vision — being prompted when I forget something important — is really, really helpful. We’re not fully there yet on the UX, and a huge part of that is the state of the models: too big, too slow, sometimes too expensive. Over time, all of that becomes less of an issue.

Looking beyond leasing, do you see Grotto’s technology evolving into a broader platform for improving high-stakes human interactions across other industries, and if so, which sectors are most compelling to you?

Right now, the winning SaaS we see is deeply vertically focused, and we’re focused on this industry. When we’re in a sales process against competitors, there’s a depth of functionality we can provide that differentiates us that’s the result of spending thousands of hours on-site with leasing teams. That’s what helps us win. To do that and keep winning, we need to stay really focused on this industry.

Thank you for the great interview, readers who wish to learn more should visit Grotto AI.