All casesNo. 078
BH Management Puts Voice AI in Front of Its Call Center
Calls bound for the contact center reach AI first; agents keep the ones that need nuance
- Case no.
- 078
- Primary source
- Funnel panel recap, customer story, press releases, trade interview
- Source published
- 2026-03-27

1. Why this case
BH Management is a US apartment company that both owns buildings and manages them for other owners. In a Funnel panel recap dated 2026-03-27, BH is described as “a national platform that spans both ownership and third-party management,” with more than 93,000 units in 42 states, No. 13 on the National Multifamily Housing Council (NMHC) Top Managers list and No. 37 on its Top Owners list. BH’s own press release (2026-02-19) says it was founded in 1993 and employs 2,300 people.
The change in this case is a single routing decision. BH’s contact center is the central team that answers renters’ calls, and the recap calls it BH’s “front door” to The Mint Experience, the company’s branded resident experience. BH put Funnel’s agentic Voice AI in front of that team, “routing all incoming calls that would normally go to the BH call center.” The stated aim is to give the contact center more room for tasks that “require more nuance.” The recap’s example of a question that needs less is “What time does the fitness center open.”
The case is useful for two reasons. First, the AI was added on top of a team that BH had already centralized. BH had moved contact center operations, along with leasing, renewals, account management, and screening, into central shared-services teams, and the Voice AI now takes the first pass at the calls coming into one of those teams. Second, BH manages buildings it does not fully own, so it has to convince owners that the centralized model is worth it. BH’s president, Melody King, said at the panel that the Voice AI’s numbers are what make that case.
The reported figures are an average of 41 hours a day saved for the contact center, an average of 492 calls a day “avoided or handled by AI without the assistance of a human,” and a +200% year-over-year increase in containment rate. They come from a vendor’s recap of a panel at the vendor’s own event. The measurement period and the containment baseline are not stated.
2. Problem
2.1 Hundreds of communities, each doing it differently
Funnel’s BH customer story describes the starting point. BH “needed to standardize processes across 300+ communities that were each doing things their own way. The industry-wide combination of rising costs, renter expectations for “Amazon-easy” experiences, and overburdened onsite teams made this change urgent.”
Christi Weinstein, BH’s COO, describes the cost to the people at each property: “We’d overly burdened our onsite teams to where the customer was no longer first, which is unfortunate because that should be the mentality onsite.”
BH’s answer was centralization. With Funnel, the customer story says, BH “centralized key functions: leasing, renewals, account management, screening, and contact center operations.” In a June 2026 interview with Multifamily Executive, CEO Joanna Zabriskie named the four pillars of the centralized services model: “Contact Center, Applicant Screening, Renewals, and Resident Account Management.”
2.2 The contact center as the front door
Centralization gave BH a central contact center, which the recap also calls “the BH call center” and BH’s “front door” to The Mint Experience. The recap frames the Voice AI change in terms of what that team should spend its time on: tasks “that require more nuance” than a question about gym hours.
2.3 Owners have to be convinced
BH is “one of the largest third-party managers in the country.” The recap notes that for a third-party manager, “gaining buy-in from various owner groups for a new operating model presents a unique challenge for operators.” The customer story puts it the same way: “BH had to prove that shared services could deliver benefits across portfolios they don’t fully control.”
King described the difficulty directly: “Centralizing in a third-party platform has been the most challenging and the most rewarding thing that we’ve done,” she said. “It’s sometimes hard to explain or prove [the value] out to a client, it’s a very hard thing, but it’s rewarding because we learn a lot and that allows us to take that insight and apply it to every asset across the board.”
Confirmed facts: BH is an owner-operator and third-party manager; 93,000+ units in 42 states (recap, 2026-03-27); 300+ communities that ran processes their own way (customer story); centralized functions include contact center operations; four pillars named by the CEO (June 2026); third-party managers need owner buy-in for a new operating model.
3. Solution
3.1 All contact-center calls go to Voice AI first
The recap states the change in one sentence, which is a fragment in the original: “After implementing Funnel’s agentic Voice AI solution, routing all incoming calls that would normally go to the BH call center.”
The scope is the calls that would have gone to the call center. The stated result is “more opportunity for the contact center” to “focus on tasks that require more nuance than, “What time does the fitness center open.""
3.2 Part of the Mint Experience and the centralized services model
The Voice AI sits inside a broader program. The recap says “Leveraging AI and the latest technology, and moving to a centralized model were core components of BH’s Mint Experience.” Zabriskie defined the Mint Experience in June 2026 as “BH’s resident-first service philosophy,” which “combines centralized support, technology, and on-site hospitality to remove friction from everyday moments.” She said “85% of our portfolio” is now on Mint Services.
Funnel is the platform underneath. Zabriskie calls it “our artificial intelligence (AI)-powered CRM partner.” The customer story lists the Funnel products BH uses as “CRM, AI, online leasing, + resident portal,” with Yardi as the property management system. It describes how AI and people share conversations across the renter journey: “Conversations that start with AI flow seamlessly to human agents, without renters ever noticing a gap,” and “Funnel’s AI automatically updates guest cards with call summaries, key details, and next steps.” A guest card is the CRM record for a prospective renter. The story describes Funnel’s AI across BH’s renter journey as a whole.
3.3 The same partner tried a resident transaction in ChatGPT
The recap opens BH’s section with a different project. In February 2026, BH and Funnel completed what both companies call the first rent payment inside ChatGPT. BH’s press release describes it as “a controlled pilot in partnership with Funnel,” and Funnel’s release calls it “Proof of concept, not a rollout.” King linked the pilot to how renters already behave: “What we recognize is that people are already operating in those digital platforms. They’re already using AI assistance, and that’s just kind of a go-to for them.”
The pilot shows BH’s wider interest in agentic AI for residents.
Confirmed facts: Funnel’s agentic Voice AI receives all incoming calls that would normally go to the BH call center; the contact center is called BH’s “front door”; AI and centralization are named as core components of the Mint Experience; 85% of the portfolio on Mint Services (June 2026); Funnel is BH’s AI-powered CRM; the ChatGPT rent payment was a controlled pilot (February 2026).
4. Impact
The figures below are from the 2026-03-27 Funnel recap. Each is attributed to BH but reported by the vendor. The recap does not state the measurement period.
| Metric | Figure | Basis |
|---|---|---|
| Containment rate | +200% year over year | Increase in the rate; the rate itself and the years compared are not stated |
| Contact center time saved | 41 hours per day | Average |
| Calls without a human | 492 calls per day | Average; calls “avoided or handled by AI without the assistance of a human” |
Containment rate generally means the share of calls an AI system finishes without passing them to a person.
The 492 figure combines two outcomes, calls “avoided” and calls “handled,” in one count. The recap does not split them.
The recap ties these numbers to the owner problem in section 2.3. Right after the passage on owner buy-in, the recap paraphrases King: the secret weapon in presenting true value is in the data. The data it cites are the 41 hours and the 492 calls.
No figures are given for caller satisfaction, wait times, cost, or contact center staffing.
5. Insight
The AI took the first pass at work that had already been centralized once. BH first centralized contact center operations as part of standardizing 300+ communities. The Voice AI is a second move on top of that: the calls that would have reached that central team now reach the AI first. A central call center made “all incoming calls that would normally go to the BH call center” a clear, routable scope. That scope would be harder to define if every community still answered its own phones.
The contact center’s job is defined by what is left. The recap does not describe new tasks for the contact center. It describes the remaining calls: those that “require more nuance” than a simple factual question. The people’s work is whatever the AI cannot close.
For a third-party manager, the AI’s call log doubles as a sales document. BH has to sell centralization to owners who do not report to it. King’s point is that the Voice AI produces numbers an owner can read, such as hours saved and calls handled per day.
The reported metrics measure the operator’s side. Hours saved and calls handled count what the contact center no longer does. They do not show whether callers got the right answer, how many hung up, or whether an AI-handled call later came back as a human call.
Limits. All three figures come from a vendor’s recap of a panel at the vendor’s event. The measurement period is not stated, the containment baseline is not given, and “avoided or handled” mixes two outcomes. The recap does not describe the contact center before the change (call volume, wait times, staffing), which call types the AI resolves, or how handoffs work, and it reports no staffing changes. Unit counts differ by source: 93,000+ in the recap and BH’s February press release, 94,000 in Funnel’s customer story and “over 94,000” in the June interview.
One-line takeaway: BH put Voice AI in front of a contact center it had already centralized, sent every call bound for that team to the AI first, and used the AI’s daily counts to show owners what the centralized model is worth.
6. Implementation hypotheses (not in the source)
6.1 Repointing the contact-center line and handing off with context
Basis: The recap says BH began “routing all incoming calls that would normally go to the BH call center” to Funnel’s Voice AI, leaving the contact center “tasks that require more nuance.” The customer story says “Conversations that start with AI flow seamlessly to human agents” and that “Funnel’s AI automatically updates guest cards with call summaries, key details, and next steps.”
The simplest way to route “all incoming calls that would normally go to the BH call center” is at the phone layer. Property numbers that already forwarded to the central contact center would be pointed at the Voice AI instead, so the change needs no new number for renters. The dialed number would tell the AI which community the caller means, and the AI would answer factual questions, such as amenity hours, from that community’s data in the CRM. When a call needs judgment or an action the AI cannot take, it would transfer the call to the contact center queue. Because Funnel’s AI already writes call summaries to the CRM record, the agent who picks up would likely see what the caller asked before saying hello. That is what would make the handoff “seamless” in the customer story’s sense.
6.2 Per-call outcome tags that roll up into owner reports
Basis: The recap reports “an average of 41 hours per day” saved and “an average of 492 calls per day avoided or handled by AI without the assistance of a human,” presented as the data for winning buy-in from “various owner groups.” The customer story says BH had to prove shared services across “portfolios they don’t fully control.”
Daily averages like these most likely come from tagging each call’s outcome in the call log: ended by the AI, transferred to an agent, or the caller’s need met another way (one plausible meaning of “avoided”). The 41 hours is probably not a timed measurement, since the saved time never happened. It is more likely an estimate built from calls that did not reach an agent and an assumed agent handling time. Because every call arrives on a property’s number, the same tags can be grouped by community and by owner. That would let BH show each owner group the calls and hours for its own buildings, which is the evidence King describes using to win buy-in.
7. Sources
- Funnel — Panel recap | From reactive AI to truly agentic systems
https://funnelleasing.com/panel-recap-from-reactive-ai-to-truly-agentic-systems/
2026-03-27 (accessed 2026-11-17) - Funnel — Faster leasing, happier renters, stronger teams (BH customer story)
https://funnelleasing.com/case-study/bh-mint-experience
Publication date: not shown on page; cites the 2026 NMHC lists (accessed 2026-11-17) - BH Management — BH Management Completes Industry’s First Rent Payment Inside ChatGPT, Advancing the Future of Resident Experience
https://livebh.com/about/newsroom/press-release/bh-management-rent-payment-inside-chatgpt/
2026-02-19 (accessed 2026-11-17) - Funnel — Funnel processes the world’s first rent payment inside ChatGPT
https://funnelleasing.com/press-release/funnel-processes-the-worlds-first-rent-payment-inside-chatgpt
2026-02-19 (accessed 2026-11-17) - Multifamily Executive (Christine Serlin) — How BH Management Is Positioning for What’s Next
https://www.multifamilyexecutive.com/leadership/how-bh-management-positioning-whats-next
2026-06-03 (accessed 2026-11-17)