AI for housing and healthcare operations
EliseAI
EliseAI builds conversational AI that automates housing and healthcare operations. Its property management platform handles leasing inquiries, tour booking, maintenance requests, payment reminders and renewals across text, email, chat and voice. EliseCRM brings renter interactions, workflows and reporting into one place. For medical practices, HealthAI automates non-clinical work: answering patient inquiries, scheduling appointments and sending billing reminders. VoiceAI answers calls and manages follow-ups, while integrations with electronic health records and other practice systems connect conversations to administrative workflows.
Founded in 2017 by Minna Song and Tony Stoyanov, EliseAI expanded from housing into healthcare in 2023. Song leads the company as CEO, with Stoyanov identified as co-founder and CTO. Its Apollo agent, introduced in 2026, gives property teams a conversational interface to the existing Elise platform. Staff can ask it to reschedule a colleague’s tours or identify units with recurring maintenance issues. Apollo uses the platform’s property data and settings and inherits each user’s existing permissions.
The business serves property owners and operators alongside healthcare practices, embedding automation in ongoing resident and patient services. In June 2026, the company reported passing $200 million in annual recurring revenue and powering one in six U.S. apartments.
Explore the recordings
This archive currently contains one EliseAI-tagged recording, so the most useful routes are thematic rather than chronological. The guide below describes what the recording presents; its technical descriptions, performance characteristics, demonstrations, and domain references are recorded claims, not confirmation of EliseAI’s current products or capabilities.
Start here: the complete voice-agent architecture
Watch Realtime Voice Agents with Frontier Intelligence — Bohan Li, EliseAI from beginning to end for the broadest route through the archive. Bohan Li presents a real-time voice-agent harness as a cascaded perception, planning, and control system, using a self-driving stack as an analogy. This is the archive’s single overview recording, not a company-history source.
Technical path: latency, corrections, and concurrency
Use the same recording as a component-level study. It describes tentative streaming transcription followed by slower contextual correction, cancellation of stale corrections, eager response generation with delayed release, background tool work that injects context, and text-to-speech prefix caching with overlap suppression. Product and provider names—including Flux, Scribe V2, and Cartesia—reflect the architecture discussed in the recording and should not be treated as evidence of today’s integrations or implementation.
Applied path: clinic-booking demonstration
For an application-focused route, go to the prerecorded clinic-booking call within the talk. The demonstration covers identity collection, insurance intake, interruptions, appointment options, and booking. The talk also briefly describes work in housing and healthcare; these are statements made in the recording, not independently verified current offerings.
1 talk
Newest first
1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.