Sierra co-founder Clay Bavor joined Funnel CEO Tyler Christiansen at Forum to discuss the future of enterprise AI and the way work gets done.
For most of its history, multifamily adopted new technology after other industries did. Digital payments, CRM, automation, and digital leasing all took longer to gain traction here than they did elsewhere. AI is changing that pattern.
At Forum 2026, Clay Bavor, co-founder of Sierra, joined Funnel CEO Tyler Christiansen for a conversation that landed on a bigger point than product features or demos. For the first time, multifamily is operating much closer to the edge of what AI can actually do. As Christiansen put it, “The technology that is now in the hands of your teams and your renters is literally being deployed by SiriusXM, maybe a few weeks before.”
That matters because it gives multifamily a rare chance to learn from the frontier instead of waiting for the market to mature around it. Bavor brings a useful vantage point to that conversation. He helped build key products at Google, and now leads one of the fastest-growing AI companies in the market.
Most importantly, he understands where AI is already creating measurable business value across customer-facing industries. As he said, “The pace at which all of this is unfolding is kind of extraordinary.” He then narrowed the conversation to the two areas where AI is already having a clear impact on the economy: coding and customer-facing work.
That second category should have the full attention of multifamily leaders.
The atomic unit of AI and automation is the workflow or the process
One of the most useful lines from the conversation came early. Bavor said, “The atomic unit of automation is the workflow or the process.” He followed it with the larger point: “Jobs are a composite of many different processes.”
The real AI shift comes from breaking work into its component parts, then identifying which parts show up at high volume, follow repeatable patterns, create friction, and slow teams down across the renter journey.
In multifamily, that means the opportunity sits inside the work itself: searching inventory, scheduling tours, creating prospects, answering common questions, and moving prospective renters through the early stages of leasing with more speed and consistency. Those workflows carry weight because they happen constantly, require follow-through, and shape both team capacity and the renter experience.
That is where this moves beyond the standard chatbot era. A system that can take action inside the workflow carries much more weight. Over the last 12 months, Funnel’s AI leasing agent, powered by Sierra, handled 1,000,000 conversations, created 161,000 prospects, scheduled 30,000 tours, and maintained roughly an 86% containment rate. Around 60% of those conversations focused on apartment search and around 30% focused on scheduling tours. This is not marginal activity sitting off to the side. It is real leasing work moving through the system.
The work that rises to the top is the work that happens constantly, carries real operational weight, and can move faster without introducing new chaos. “A juice to squeeze ratio on those workflows,” said Bavor.
AI’s impact is stronger when workflows connect
In the traditional 1:100 property-centric model, the onsite team owns almost everything. Leasing, applications, renewals, resident experience, communication, follow-up, and administrative work all sit at the property level, often with the same small group of generalists juggling all of it at once.
Teams have to context-switch constantly. Priorities compete with each other. Follow-through varies depending on who is available, what else is happening at the property, and which task wins attention in the moment. The renter feels that as inconsistency across the journey: a fast response in one moment, a delay in the next, a dropped handoff, a repeated question, a different answer from a different person.
AI does not fix that on its own. In a property-centric model, it can help onsite teams move faster, but the structure of the work stays the same. The same team still juggles too many responsibilities. The same friction points still exist. The same operational strain still sits at the property.
Centralization and role specialization start to change that. They pull high-volume, repeatable work out of the catch-all onsite role and into workflows that can be handled with more consistency, greater expertise, and better scale. That shift improves the experience for teams and renters alike, but it also raises the stakes on connection. Once work moves across specialized teams, the handoffs matter more. Context has to carry across the journey. That is where AI starts to create more meaningful value. It can help connect the workflows, reduce friction between steps, and make the renter experience feel continuous instead of dependent on how much one onsite team can hold together at once.
Bavor described the downside of disconnected workflows directly: “If you think about those separately, you end up with this kaleidoscopic fragmented view of the resident and them having a fragmented experience.”
He explained that Sierra has invested in “not just broadening the set of workflows and applications that we automate, but connecting the dots between those.” He described long-term memory as “the ability for an agent to remember interactions with customers across different sessions, connect the dots on those.” He also described a higher-level architecture in which “a super agent” can include “a sales, a booking, a support, a maintenance agent,” then “put on the appropriate hat, recall the relevant memories, and then get to solving the resident’s problem.”
For multifamily operators, that continuity shapes both the renter experience and the operating model behind it. When workflows connect, service gets more consistent, teams get more leverage, and AI starts to strengthen the model instead of just helping an overloaded onsite team work a little faster in a siloed system.
Real ROI starts with real work
Bavor pushed back on the idea that AI pilots rarely turn into real business value. At Sierra, he said, “somewhere between 90 and 95% of all of our proofs of concept have converted to long-term paying contracts.” Sierra prices against completed work, not AI activity. That keeps the conversation anchored to outcomes instead of experimentation for its own sake. It also forces a much harder standard. The work needs completed, the task has to carry value, and the result must hold up under real operating conditions. That leaves less room for what Bavor called “AI tourism” and a much higher bar for what counts as progress.
That is also what separates serious deployments from the wave of AI noise flooding the market. Multifamily does not need more demos, more vague promises, or more vendors performing fluency in AI. It needs systems that complete useful work, improve service, strengthen economics, and keep working once the spotlight is gone.
In a regulated industry, the invisible work matters most
Multifamily does not have much use for AI that sounds smart and gets the details wrong.
This is an industry where accuracy, consistency, and control carry real operational weight. Leasing, pricing, fee transparency, policies, and resident communication do not leave much room for improvisation.
Bavor said it directly: “It’s just totally unacceptable for an AI agent to make up stuff, pricing fees in particular, when you’re legally required to get it right.”
That is why the engineering underneath the experience matters so much. Sierra breaks the task into parts, uses different models for different jobs, adds oversight to check accuracy, and pulls critical information like fees from fixed sources. The experience can still feel conversational, while tighter controls sit underneath where accuracy matters most.
The operators who build the right foundation have the advantage
Building a compelling demo now takes very little. Building something that survives real operating conditions takes much more.
The hard part sits below the surface: release management, rollback, latency, oversight, edge cases, and the long list of details that determine whether a system actually holds up once it goes live. “The AI agent iceberg goes deep, it goes really, really deep,” Bavor said. Like an iceberg, the visible part is the conversation. The hidden part determines whether the system can survive real operating conditions. Under the surface sit all the things a demo never has to prove: how the system handles latency in a live conversation, how it recovers when something breaks, how it rolls back safely, how it manages exceptions, how it applies guardrails, how it gets audited, how it improves over time, and how it holds up across thousands of real interactions instead of one controlled example.
For multifamily, the takeaway is straightforward. This industry no longer needs to watch from the sidelines and wait for someone else to prove what AI can do. It is already close enough to the frontier to learn from it directly.
The operators who move forward from here with stronger workflows, cleaner data, more connected systems, and clearer ownership across the renter journey will be in the best position to turn AI from interest into advantage. As Bavor told the room, “you are all pioneers and truly among the first in the world to deploy these advanced AI agents.” For once, multifamily is not watching the future arrive from the sidelines. It is helping shape it.
