All casesNo. 077
Greystar Puts Leasing AI Inside the CRM
AI takes first contact, follow-up, and tour booking; onsite teams focus on conversion
- Case no.
- 077
- Primary source
- Funnel panel recap (vendor blog)
- Source published
- 2026-03-27

1. Why this case
Greystar ranks No. 1 on both the NMHC Top Owners and Top Managers lists, the US apartment industry’s rankings of owners and managers, with more than one million units managed. At Funnel’s Forum conference in Scottsdale, Arizona, the leasing software vendor ran a panel on moving “from reactive AI to agentic systems.” Melody (Mel) Reid, Managing Director of Property Systems Engagement at Greystar, described what changed in Greystar’s leasing work.
The case is useful because Greystar’s change did not start with a new AI feature. In Funnel’s account, “the shift to agentic AI started with infrastructure, not automation.” After testing bolt-on tools, Greystar concluded that AI has to sit inside the system of record, the CRM where leasing teams already keep each prospect’s record and conversation, rather than around it. Once the AI was inside the CRM, the leasing work was divided. AI handles first contact, follow-up, and tour scheduling. Onsite teams focus on conversion and the renter’s experience. A second mode runs alongside: AI drafts a personalized reply, and a leasing team member reviews it and sends it.
Funnel reports that across Greystar’s portfolio in 2025, AI supported 3.5 million messages sent, with a 62% containment rate, and scheduled 94,000 tours across 1,400 Greystar properties. Leasing teams adopted 60% of AI-suggested responses and sent more than 500,000 AI-assisted responses in 10 months. All figures come from a recap written by the vendor and are not independently audited.
2. Problem
The Greystar section of the recap describes three pressures.
- Scale. Reid’s core challenge, in Funnel’s words, is “ensuring the systems Greystar deploys can actually scale across an enormous platform.” Anything that works at one property has to work across a portfolio of more than one million units.
- Bolt-on AI did not deliver. “After testing other bolt-on solutions, they found that AI only delivers real value when it is embedded directly into the system of record, not bolted on around it.”
- Data spread across systems. The recap puts it as “When information lives in silos, AI loses the context it needs to be effective.” Reid said: “More is not better with AI. Less is much better towards different disparate technology systems because what makes AI smarter is data in context.” She added: “Clean data equals smarter AI. So when you have a lot of disparate systems they’re all defining things differently or it’s really hard to understand.”
Reid’s own framing points to a consistency problem in leasing itself: every inquiry needs an answer, a follow-up, and ideally a tour. She frames the 3.5 million AI-supported messages as “three and a half million times someone didn’t drop the ball,” and says, “I’m not even sure all of our teams would have accomplished all of that manually on their own in the past.”
Confirmed facts: No. 1 on NMHC Top Owners and Top Managers lists; more than one million units managed; Reid’s focus is systems that scale across the platform; bolt-on AI tools were tested first; AI judged to deliver value only inside the system of record; disparate systems define things differently.
3. Solution
3.1 AI inside the system of record
Greystar put Funnel’s AI and renter-centric CRM in place. According to the recap, “Greystar restructured how leasing workflows operate around a single source of truth.” Reid calls this a key lesson: “AI native to CRM has been really, I think, a key lesson learned for us.” She added: “It’s really how they apply the LLMs to the technology to become truly agentic that again I think is going to be the needle mover.”
In practice, the AI works from the same record the leasing team uses, instead of from a separate tool attached around the CRM.
3.2 What AI does on its own
The recap states the division plainly: “AI now handles first contact, follow-up, and scheduling, creating space for onsite teams to focus where they add the most value: conversion and experience.”
Three parts of the leasing funnel are named:
- First contact. AI answers new inquiries. Many of the 3.5 million messages were sent “after hours,” when the leasing office is closed.
- Follow-up. AI follows up with prospects after the first exchange.
- Tour scheduling. Funnel describes “agentic AI tour scheduling, which covers a pivotal moment in the funnel.”
Funnel measures this mode by containment: the share of interactions “where AI fully resolves the interaction.”
3.3 What AI drafts and people send
The second mode keeps a person in the loop. “By pulling in guest card data, communication history, and renter preferences, Funnel’s AI generates personalized messages that leasing teams can review and send.” A guest card is the multifamily industry’s term for a prospect’s record: contact details, what they are looking for, and their history with the property.
Funnel describes the result as “blending speed with context instead of sacrificing one for the other.” Reid describes the division with a golf image: “If we’re doing our jobs right, the AI just walks up and puts the ball on the tee, and then your team walks up and hits it home.”
3.4 What the onsite team does now
With first contact, follow-up, and scheduling moved to AI, onsite teams work on conversion, turning prospects into signed leases, and on the renter’s experience. Reid ties the tour to the lease with a general industry statistic, not a Greystar result: “A lot of statistics say if a person comes to tour, they’re 50% more likely to lease.” She continued: “Leveraging AI to get those tours scheduled more consistently is huge, and it’s driving more foot traffic to our properties.”
Confirmed facts: Funnel AI and renter-centric CRM; leasing workflows restructured around a single source of truth; AI handles first contact, follow-up, and scheduling; onsite teams focus on conversion and experience; AI-suggested responses drawn from guest card data, communication history, and renter preferences, reviewed and sent by leasing teams.
4. Impact
Figures from the Funnel recap published 2026-03-27.
| Metric | Figure | Basis |
|---|---|---|
| Messages sent with AI support | 3.5 million, “many after hours” | Greystar portfolio, 2025 |
| Containment rate | 62% | Interactions AI fully resolves; listed with the 2025 portfolio results |
| Adoption of AI-suggested responses | 60% | Period not stated |
| AI-assisted responses sent by Greystar teams | 500,000+ | “In just 10 months” |
| Tours scheduled by AI | 94,000 | 1,400 Greystar properties, 2025 |
Reid on the message count: “I mean, the three and a half million messages sent, to me, what this number represents is not just a volume gain, but it’s the value. It’s presenting back to our assets. This is three and a half million times someone didn’t drop the ball.”
On engagement, she said: “Prospects are being nurtured consistently, and more personalized responses are going to our prospects and that’s driving more engagement. It’s been really beneficial… and I’m not even sure all of our teams would have accomplished all of that manually on their own in the past.”
Funnel says tour scheduling is “increasing conversions,” and Reid says it is “driving more foot traffic,” but no conversion, lease, or foot-traffic figure is given.
5. Insight
Infrastructure came before automation. Greystar’s lesson was not that one AI product beat another. It was that AI bolted onto the side of the CRM could not use the data it needed, and that AI inside the system of record could. Reid’s point that “Clean data equals smarter AI” makes fewer systems with shared definitions a precondition for the AI work, not a cleanup project for later.
Two modes, two measures. Greystar runs AI in two ways. Where AI acts alone, the measure is containment: 62% of interactions fully resolved. Where AI drafts and a person sends, the measure is adoption: 60% of suggested responses used, and more than 500,000 sent in 10 months. Each measure answers a different question. Containment shows how much work AI finishes. Adoption shows how often people accept what AI proposes.
The tour is the handoff. In Reid’s golf image, AI puts the ball on the tee and the team hits it. In the leasing funnel, the tee is the booked tour. Before it, AI sends the first reply and the follow-ups and books the tour, and it drafts the replies that leasing staff review and send. After it, the visit and the lease are the onsite team’s work.
Consistency is the claimed value, not volume. Reid reads the 3.5 million messages as times “someone didn’t drop the ball.” The value she describes is that prospects “are being nurtured consistently,” including after hours, not that the team sends more messages.
Limits. The only source is a recap written by the vendor, and all figures are Funnel’s. The recap gives no lease-conversion, occupancy, staffing, or cost figures for Greystar. The 50% tour-to-lease figure is Reid citing general statistics. The 1,400 properties are where AI scheduled tours in 2025, and the recap does not say what share of the portfolio runs the AI. The periods differ: 2025 for messages and tours, 10 months for AI-assisted responses, and no period for the 60% adoption rate.
One-line takeaway: At Greystar’s scale, putting the AI inside the CRM let it take over first contact, follow-up, and tour booking, and let leasing teams send AI-drafted replies after review, with the booked tour as the point where people take over.
6. Implementation hypotheses (not in the source)
6.1 Suggested replies drafted from the guest card and approved in the same thread
Basis: “By pulling in guest card data, communication history, and renter preferences, Funnel’s AI generates personalized messages that leasing teams can review and send,” with a 60% adoption rate.
Because the AI lives in the CRM, the draft most likely appears inside the prospect’s conversation thread, next to the guest card, rather than in a separate tool. The leasing agent would see the draft with the prospect’s stated move-in date, unit preferences, and earlier messages, and could send it, edit it, or discard it and write their own. An adoption rate implies the system records which of these happened for every suggestion. That log is also what would tell Greystar where drafts fail: a property with low adoption would point to missing or inconsistent guest card fields, which fits Reid’s point that with disparate systems “they’re all defining things differently.”
6.2 Containment with the conversation staying on one record
Basis: “AI now handles first contact, follow-up, and scheduling,” with a “62% containment rate, where AI fully resolves the interaction,” and many messages sent “after hours.”
A 62% containment rate means the remaining interactions were not fully resolved by AI. The likely design is that AI answers every inbound message first, at any hour, and when it cannot resolve a question it flags the conversation for the property’s leasing team instead of passing it to a separate queue. Because the conversation and the guest card are the same record, the person picks up with the full history rather than asking the prospect to repeat themselves. Containment would then be counted per conversation as the share closed without a human message. The after-hours messages suggest the AI’s replies go out immediately, while flagged conversations wait for office hours.
6.3 Tour booking against each property’s calendar in the CRM
Basis: “In 2025, Funnel’s AI scheduled 94,000 tours across 1,400 Greystar properties,” and leasing workflows were restructured “around a single source of truth.”
To book tours at 1,400 properties, the AI has to know each property’s tour availability. The likely source is the same CRM, where each property keeps its tour slots and agent schedules. When a prospect agrees to a time, the AI would write the appointment into the property’s calendar and onto the guest card. The onsite team would then see the booked tour together with the conversation that led to it. This is the handoff in Reid’s golf image, the ball placed on the tee, and it only works if the booking and the prospect’s history sit in one place.
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)