Clay Bavor — Sierra — Future of agentic AI

Recorded live at Forum, this episode features a rare conversation with one of tech’s most influential minds: Clay Bavor, former head of Google Labs and now co-founder of Sierra.

Clay joins Funnel CEO Tyler Christiansen to talk about the future of agentic AI—intelligent systems that don’t just respond, but act. Together, they unpack why this moment in AI is so significant, where the real impact lies, and how AI can elevate customer experiences by removing the rote and making room for what’s truly human.

They also explore:

Funnel is proud to partner with Sierra to bring the best conversational AI to multifamily. 

Whether you’re shaping your AI strategy or just trying to understand what’s next, this conversation is packed with clarity, insight, and inspiration.

 

 

Don’t miss more conversations like this at Forum 2026. Register today!

 

Episode transcript:

0:00 Forum highlight — Clay Bavor + the future of agentic AI

Clay Bavor: There is an opportunity to help the great companies of the world—the great businesses of the world—build better, even more human experiences using AI across all parts of the customer journey. 

The Fenix agent actually just scheduled an appointment in Mandarin, which is pretty cool. My guess is most of you don’t staff Mandarin contact center agents around the clock. 

We want to have folks doing much higher-value things—make outbound sales calls, be on higher-value upsell or cross-sell calls. So imagine an AI in your case schedules the appointment and handles some basic questions, but someone who would’ve been doing the scheduling is instead making an outbound call: “Hey, I noticed you’re joining us. We actually have wine coolers. Do you need an extra parking space?” That’s revenue. That’s great. That is so much more valuable than scheduling the 73rd appointment for that day. 

I think the impact of all of that over time is likely under-hyped.

1:03 Introduction

Alex Howe: Welcome back to multifamily unpacked. This was recorded live at Forum, and I’m not exaggerating when I say it was one of the best-received conversations. Our CEO Tyler sits down for a conversation with Clay Bavor, the former head of Google Labs and co-founder of Sierra AI.

Clay’s not just shaping the future of AI—he’s helping build it. From leading Google’s most ambitious AI efforts to now, pioneering agentic AI, this guy’s résumé reads like a tech history lesson. And in this session, he breaks down what this moment means for AI, why the hype is both real and overblown, and how AI can create more human customer experiences, not less.

We’re proud to partner with Clay and the Sierra team to bring this tech to multifamily through our AI solutions — Fenix. 

Let’s get into it.

1:55 Welcome Clay Bavor, co-founder of Sierra AI

Tyler Christiansen: Alright, well hopefully you guys are as excited about AI as we are and all the new cool things that’ll be coming. We were incredibly grateful for this opportunity to speak with Clay, who I’ve gotten to know this morning even better. 

So, Clay, the first question that I’ve wondered all along with Sierra—between your background leading Google Labs, the 18 years you spent there, your co-founder Bret Taylor on the board, chairman of OpenAI, former CEO of Salesforce—why now and why in particular?

Just a note to brag a little bit more on Clay if we haven’t enough: Sierra was recently named one of the 50 most innovative AI companies in the world by Forbes magazine—they were on the Forbes 50 list. And what I noticed on that list is there’s a description of what the companies do and only one that was customer service software was Sierra. Right. So why now and why Sierra?

2:51 Why now, why customer service?

Clay Bavor: First of all, Tyler, thank you so much for having me and letting me be part of this event. It’s been terrific to meet many of you.

The question “why now” starts with what’s happening in technology and AI at this moment, and I’ve been obsessed with computers all of my life. Over the past five years I feel like I’ve been seeing science-fiction books come alive in the real world. When I was at Google several years ago, I remember we were poking at one of the early large language models—the kind of AI model that powers ChatGPT—and one of the surest signs of intelligence is the ability to think and reason and use metaphor. We asked this AI model, “Please describe the 2008 financial crisis using movie references.” Without skipping a beat, it said the 2008 financial crisis was a lot like the movie Inception, except instead of dreams within dreams, it was debt within debt.

We were like, whoa. Okay—that’s really smart. I wouldn’t have thought to put it that way. So what computers—what software—all of a sudden has become able to do: have conversations, think, reason, decide, seemed to Bret, my co-founder, and me seemed nothing short of extraordinary and, in particular, extraordinary material with which to build a new product and business.

So why customer service in particular? We think about it much more broadly than that, but to be concrete, what we’re trying to resolve is this age-old tension between wanting to provide great customer service—and more broadly a great customer experience—and the cost of doing that. Many companies hide away the 800 number: “Please do anything but call us.” There are websites devoted to helping consumers find 800 numbers to get ahold of a company. And if you do, you’re being told to get off the phone as quickly as you can.

I was walking into work the other day behind a gentleman who was clearly on a phone call on his AirPods. He starts shaking and saying, “Representative.” I was like, bro, I’m trying to help with that.

So to come back to your question, Tyler, we think there is an opportunity to help the great companies of the world build better—and even more human—experiences using AI across all parts of the customer journey. 

Yes, service and support, but also, in this context, if I’m a potential renter and there are three or four different floor plans, and I have a family, which one would work best for me given these circumstances? I have a pet—well, it’s actually a pet iguana. Would that work? And so getting advice and recommendations and help, just as you would from a concierge. That’s the kind of experience we want to deliver to our customers’ customers. The technology behind it is the seed crystal of “why now.”

5:55 Sierra aesthetic (and iguanas)

Tyler Christiansen: love that. And by the way, one of the things I’m super impressed about with Clay is his genuine interest in this industry. We sat down for breakfast this morning and he’d done all his research. It seems like you’ve done research on the challenges we all deal with, because the pet iguana thing is very real. 

Clay Bavor: Are iguanas a thing? Are they common? I’m imagining the crazy questions you must all get.

Tyler Christiansen: One thing I also want to highlight that we’ve been talking about all morning is: when used properly, AI actually makes the service more human. One thing I mentioned in my keynote—Funnel thinks a lot about design, both in our logos… I’m going to put you on the spot, because I’ve heard this on one of your podcasts. The reason why you picked “Sierra” as a name and even the aesthetic of your office—could you speak to that a little bit, the ethos?

6:40 Technology to create human experiences

Clay Bavor: Oh, I love this question. Thanks for asking. We wanted the name of our company to reflect the natural world, have humanistic overtones. In contrast, there’s a company that does voice agents called “Replicant”—like, aren’t those the things that killed people in Blade Runner? That doesn’t sound great. Both Bret and I grew up in California. We grew up in the Sierra mountains. It evokes images of nature, and we wanted to use technology to create better, more human experiences.

You might think AI and automation, what does that mean? The way we think about it is: in the context of a retailer, checking someone’s order status and their FedEx delivery number for the 73rd time that day is not additive. In your world, spending five or seven minutes wrangling whatever scheduling system you’re using to schedule an appointment is not additive. Making a human connection or taking time to lead a guided tour and share anecdotes and love of the property, maybe making connections with other families or renters there, that’s useful.

So we think of AI as another tool you have to optimize how you’re delivering an amazing experience to your renters by soaking up a lot of the rote routine stuff so that the amazing people who are part of your companies can do much higher-value, more rewarding, more interesting, high-touch things. We think that balance can and should shift, and we’re excited to be part of it.

8:33 Customer service across multiple verticals

Tyler Christiansen: Amazing. You’re a couple years into this journey with Sierra now. Looking back over the last 18–24 months—and would love to share some of the anecdotes of success you’ve had with your partners—what has surprised you in being a customer service agent across multiple verticals? In things like Sirius and WeightWatchers and Chubbies shorts—what are some lessons about the domain expertise you need versus where you can partner with the business to take advantage of what they know about their customer?

9:04 Nuance of different industries + multifamily partnership

Clay Bavor: One of the things I’ve loved about building Sierra is the privilege to partner with some of the great companies in the world and see inside the business. Every business across industries is very different. Tyler, you were nice to reflect that I knew something about this industry, I think I know 1% of 1% of 1%. Thanks to Cameron’s video on centralization—it was a great video—I know what that means.

It’s been rewarding and fun, and also hard. In the context of WeightWatchers or SiriusXM, how do we help them better manage subscription churn? When a subscriber calls in and is thinking, “This isn’t working for me; I’m not getting the value,” how do we make a connection? How do we do value discovery? “What are you into? Oh, you listen to college sports?” Earning the right to say, “We’ve got this great package for you that just has sports. It’s lower cost, would this work for you?”

There are nuances. With Chubbies, you mentioned Chubbies, they make very short shorts. I’m not cool enough to—or have nice enough legs to—own a pair. They have a pretty irreverent, funny brand, and they wanted to imbue their AI with a lot of persona and vibe. They created this whole persona of an imaginary guy named “Duncan Smothers.” Duncan cracks jokes; he’s got this irreverent sense of humor. They were focused on “Where is my order,” returns, exchange, but also making sure the brand and spirit come through.

One of the reasons I’m so grateful for our partnership, Tyler, is this is an industry that is very complex and deep, as you all know, and we think Sierra and AI could make a huge difference. We also have the humility to know that we don’t know enough to do it well. By partnering with Funnel and powering Fenix, it’s an opportunity for us to reach a whole additional industry that we otherwise wouldn’t be able to and bring all the goodness of what we built there. It’s kind of one plus one equals, I don’t know if it’s three or four or five, but it’s a number a lot bigger than two.

I really admire that you and the team have been clear-eyed about: we know how to do this set of things best—incredible SaaS and workflow applications you’ve built—and then to partner with someone in conversational AI. You can bring the best of all these worlds in one package, and we’re grateful to be a part of it.

11:45 Embracing AI at large scale

Tyler Christiansen: Thank you so much. We are humbled. It was a moment for us in time,again, I showed some slides from back in 2019. The idea that we’d have somebody of your caliber, Clay, and an organization like Sierra willing to partner with us is very humbling—just as humbling as having all of you as our partners.

We highlighted some success stories. One thing I will highlight, and I appreciate you saying this, I talked earlier about the balancing act of AI and humans. We want to underscore there are huge—just like centralization. One of our partners, Camden, proudly displayed on their earnings reports that centralization saved them $4 million a year. We are here for the cost savings. This industry is nuanced, complicated, and regulated, but it has historically struggled with staffing and with keeping properties staffed. Then when you do err on the side of centralize, try contact centers, then you you walk into the challenge of staffing contact centers. There’s overhead, including refilling the seats that keep opening up.

In the context of centralized contact centers in particular—you have a lot of experience with partners—what are some wins where you’ve been able to reduce staffing necessary, and on the flip side, what are some technical challenges? As folks here think about embracing AI at large scale, what are the things that, if this happens or we don’t have access to this data, it’s going to go wrong? Maybe some wins and some challenges.

13:10 AI that produces results

Clay Bavor: On the wins front, first of all, typically our agents in their steady state—after some amount of coaching and teaching—are able to resolve somewhere between 50% and 90% of all incoming questions. We’re early days with Team Essex and several others of you, so it’s hard to know exactly where that will converge in the prospect support case and so on, but I’m optimistic. 

At about the 70% level, we’ve seen companies bring down contact center hours or headcount by about 50%.

What’s interesting is many companies will bank those savings. Others will keep those people in seat but have folks doing much higher-value things. Make outbound sales calls, be on higher-value upsell or cross-sell calls. So imagine an AI schedules the appointment and handles basic questions, but someone who would’ve been doing the scheduling is instead making an outbound call: “Hey, I noticed you’re joining us. We actually have wine coolers. Do you need an extra parking space?” Other things you can attach to the leads. That’s revenue. That’s great. That is so much more valuable than scheduling the 73rd appointment for that day. Those are big-picture results and cost savings.

There are more subtle things. Seasonality is huge for retailers.

Tyler Christiansen: In this industry, we call it leasing season.

Clay Bavor: Yeah, leasing season. I bet there’s an industry more seasonal than yours—one of our customers is Minted. If any of you make holiday cards with Minted, we do every year. Very seasonal. There’s a spike when everyone realizes, “Oh my God, it’s Thanksgiving. That means I need to get my holiday cards out.” For businesses like Minted and retailers where Black Friday/Cyber Monday means that kind of spike, retailers typically have to double staffing in Black Friday/Cyber Monday. That staffing for those two weeks can be as much as a third of their total annual cost of service—hiring up ahead, paying overage, and so on.

I’m super proud: our retailers this year—one of them, whereas in past years had to double staffing—this past Black Friday/Cyber Monday added two people company-wide. We were really excited about that. We blunted that out for Minted.

And then finally, the Fenix agent actually just scheduled an appointment in Mandarin, which is pretty cool. My guess is most of you don’t staff Mandarin contact center agents around the clock. Whether it’s Mandarin, Chinese, French—our agents can speak 20 different languages, and it’s neat to meet a potential renter where they are, speak with them in their native language. 

Imagine the delight of someone moving from out of country—potentially someone Spanish-speaking—being able to speak in their native language, even in their native dialect, and have that be utterly seamless. That’s what we mean by more human experiences.

16:18 Outcome based pricing

Tyler Christiansen: That’s incredible. A lot of our partners, and to emphasize our commitment to AI, what we’re talking about, this cutting-edge technology, is available to every Funnel VLA customer. We’re going to eat that cost; we’re going to partner. On the cost side, we really love that Sierra’s challenging, it’s outcome-based pricing.

Clay helped build and manage Google Workspace years ago. It’s a SaaS product, you subscribe to that. His co-founder Bret Taylor was the co-CEO of Salesforce, the OG of CRM. Talk to us about that adoption in your business of moving toward outcome-based pricing. 

Everything you described—the value is only paid when there’s a tangible outcome. How has that shift been for you and Bret as you’ve shifted from the old-school SaaS world to outcome-based pricing? Any lessons?

17:29 Pricing incentives: win-win

Clay Bavor: The evolution of how software has been purchased is interesting. Twenty-five years ago, you’d go to Fry’s Electronics and buy a box with CDs and install it on your computer. You paid $600 for Photoshop regardless of how much you used it. Then there was SaaS, you’re renting software. Companies like Snowflake introduced consumption-based pricing, AWS as well, you pay depending on how much you use it.

Our view is: you should think of AI and AI agents as software you are hiring to do a job—and do it well, to get something done. Most tools today help a person be slightly better at getting their job done. AI, and agents in particular that can take action, can fully complete tasks: sending a satellite signal, drop-shipping shoes, sending a new battery or a DT alarm sticker in the mail, scheduling an appointment to view a property. It’s software that can fully complete tasks.

What we liked in our approach to pricing was deeply aligning our incentives with our customers. You’re in essence hiring a Sierra agent to get a job done and done well, and only when it does that should you pay. If the agent says, “I don’t know about that; let me hand you off to someone,” you haven’t realized any value and shouldn’t pay. That’s what we mean by outcome-based pricing. The fundamental insight is incentives alignment: you save a bunch of costs, what would’ve cost $10 or $20 in a contact center or keeping someone staffed to man a phone, we resolve for a fraction of that. You see huge savings, and we win when you win.

19:17 AI validation in communication

Tyler Christiansen: Phenomenal. To iterate again, the Sierra partnership is embedded in the VLA. We will be experimenting with outcome-based pricing. I talked about renewals—there are things we want to try, but we also understand your business model. My dad’s an asset manager, he likes that fixed-cost model. We are excited about this, but we want your feedback as we introduce these new workflows.

Another thing we’d love to experiment with that Sierra’s doing is authentication—validating that the consumer is who they say they are. Historically in multifamily, if I’m making a rent payment or submitting a work order, I do that when I’m logged in. In the rush to introduce AI, my candid opinion is we’re probably doing some things we might want to authenticate—like reminding people to pay rent and sign renewal offers. Maybe speak to—WeightWatchers is an example—where the agent is deployed with validation that the person is who you think you’re communicating with.

20:21 AI custom to renter needs

Clay Bavor: We think great AI agents, like the best salespeople or support people, should remember what you ordered, pull up information, and have context on you. In the WeightWatchers case, and the majority of our customers, we authenticate the user either by having the customer we work with pass in an authentication token (This is Bob Smith from such and such) or we can do fuzzy matching: name, ZIP code, last four digits of the credit card you made the deposit with, things like that. Okay, I’ve pulled up everything, and now I can deeply personalize this session and answer questions specific to you: When is your next payment due? If I’m late by this much, what’s the penalty? Does my property come with one or two parking spaces?

We think there’s huge power in that. In supporting renters, you can create this concierge-like experience. Authentication and knowing the details of the renter—their experience and products—are a really important ingredient.

21:38 Trust in AI is critical

Tyler Christiansen: That’s music to our ears. The idea of having a single renter guest card across multiple communities—and using that to personalize the journey—something we hear from our friends at Greystar all the time—is a huge opportunity we don’t do well today. 

Every year Salesforce puts out the State of the Connected Consumer, and the number one thing consumers want is to be known. I do not want to re-explain who I am. Nikki told her story about explaining who she was again. Trust in AI is critical. Once you get that validation—”Oh, hey Tyler, I see you have a flight coming up”—my blood pressure goes down immediately. “Good, I don’t have to explain that. I just want to change my flight.”

Clay Bavor: I think some circle of hell is spelling your last name for the fourth time, 12 letters. I was impressed you pronounced mine correctly, difficult last name knows difficult last names. I appreciate it. The number one predictor of a high customer satisfaction interaction in service is low effort. People talk about delight and surprise, sure, but in Maslow’s hierarchy of experiential needs: could you please just solve my problem quickly and fully, without me having to spell my last name for the fifth time? Authentication and knowing who the renter is short-circuits so much of that. You feel known; you feel seen; effort drops.

23:08 What in AI is overhyped, and underhyped?

Tyler Christiansen: Fantastic. We’ll wrap up with the speed round with Clay. This has been phenomenal. First one: in the world of AI, what is over-hyped and what may be under-hyped?

Clay Bavor: There’s a saying that with new technologies, people overestimate impact in the short run and underestimate impact in the long run. AI is no different. What’s over-hyped in AI? AI is over-hyped in AI. One of my favorite things is using the Internet Archive’s Wayback Machine and looking at businesses’ websites, when they slapped “AI” on their homepage. Whatever they were doing, now it’s “with AI,” and more recently “with AI agents.” Agents are all anyone wants to talk about.

To be more concrete: it’s naïve to say, AI will solve all known problems. We have this application built over 10 years—they say, ”Couldn’t AI do that?” Over some time period, maybe. But systems and tools like Salesforce, like Funnel, that manage all the complexities of renter lifecycle management, overseeing multiple properties, centralization, leaning on franchises, aggregating information—that’s not something you just point some AI at. There’s so much know-how about the industry and best practices imbued in those tools that those tools plus AI are going to continue to be the name of the game for a long time.

What’s under-hyped? Also AI. Over the last year, in each of your offices, a PhD in every field of human knowledge has shown up and is waiting in the corner ready to do research for you. If you’ve used OpenAI or Google’s deep research: these are AIs that can go do what would’ve been a week-long research project for a PhD and come back with a briefing. You now have 10 or 12 PhDs sitting in your office. For technical folks, you have a stable of software engineers in AI-assisted coding agents—Cursor, Codeium, and others—waiting to write whatever application you would like.

It is hard to, to predict just where that goes. But this idea that suddenly every one of us has at our disposable tens, even hundreds of kind of thinking machines that can go out and research and code and write brainstorm with us is extraordinary. 

And I think it’s just as, I think at the advent of the internet, no one could have predicted YouTube or TikTok or Uber or DoorDash. We’re at the very beginning of this next phase, and I think it is so exciting and a great gift. A to be alive at all, but b, to be alive at this point in time as these technologies come to fruition. 

And I think with such great and real and direct opportunities to apply it in, in everything that you do and in elevating the renter experience and elevating the customer experience in providing this kind of concierge like experience from, from soup to nuts and. Again I think delivering this more human experience by soaking up the ROT stuff and enabling us all to do what we do best. So I’m super excited about that and I think the impact of all of that over time is likely under hyped.

Tyler Christiansen: You know what, we’re going to end on that. That was the best possible answer I could have asked for. Thank you so much, Clay, for being here. We’re going to welcome up in a minute the centralization panel, but just—everybody here is a Funnel customer using our virtual leasing assistant. We are so excited. I couldn’t be more humbled, but excited to have a partner like Sierra for this. Please test the product, give us your feedback, because the future’s very bright.

Clay Bavor: Thank you, Tyler. Thank you everyone.

Alex Howe: Thanks for tuning in. If this episode got you thinking, here’s the next step. Join us at the Forum—the best conference in multifamily—March 23rd to 26th at the JW Marriott Camelback Inn in Scottsdale. I can assure you Forum is not the average conference. It’s built for multifamily executives who are rethinking how teams, technology, and operations come together to deliver standout renter and team experiences.

If that sounds like you, this is where you need to be. Registration’s open now at the link in the show notes. Early-bird pricing runs through November 2nd, and a heads up: Forum sells out every year. Don’t wait. Grab your seat at the best rate today. We’ll see you in Scottsdale.