May 08, 2026
RingCentral to AI Purists: You Still Need Humans and We’ve Got the Platform to Prove It
RingCentral ended the Q1 2026 with more than 11,800 paying customers for its AI Receptionist (AIR) product, a 40% increase from the previous quarter, as the company doubles down on what CEO Vlad Shmunis calls a “hybrid human-in-the-loop” approach to customer engagement.
The figure, disclosed during the company’s Q1 earnings call, represents one of the more tangible demand signals for agentic voice AI in the contact centre and unified communications space. AIR handles inbound customer enquiries over voice and text, integrates into call queues, and manages overflow and missed calls without requiring developer involvement.
The Case Against Full Automation
As agentic AI fuels speculation that fully automated customer interactions are within reach, Shmunis disagreed. He said:
“We still very much see room for a human in the loop,” adding that the future is “neither all AI nor all human, but a bit of both.”
He pointed to legal and practical constraints that will keep humans embedded in customer interactions for years to come. Healthcare was his go-to example: AI can answer billing questions, schedule appointments, and surface test results, but it cannot provide medical advice on behalf of a licensed practitioner. “For the foreseeable future, we don’t see AI being licensed as a practising physician or someone who can do prescriptions,” he said.
RingCentral’s entire RCAI portfolio is built around that hybrid mode. AIR handles the front door before a human gets involved, AI Virtual Assistant (AVA) assists the human agent in real time during the interaction, and AI Conversation Expert (ACE) analyses recordings and transcripts after the call. Insights from ACE then feed back into the system, making both AI and human agents more effective on the next interaction.
Asked by an analyst what prevents AI startups from becoming a competitive threat, Shmunis argued that they can automate interactions but can’t hand off to a human agent without third-party integrations.
“They cannot really connect to a human,” he said. “They have to integrate with third parties. At RingCentral, we’re able to serve both needs. Single platform, single invoice, single SLA.”
Contact Centre Wins and the Teams Play
RingCX, the company’s formal contact centre product, now has more than 1,700 customers, with over half using AI capabilities. A multi-location orthopaedic practice in New York cut its call abandonment rate from 22% to 8% and reduced average hold times from 30 minutes to three after deploying RingCX alongside ACE Quality Management.
The Customer Engagement Bundle (CEB), which brings informal contact centre capabilities, including contact centre-grade call queues and shared SMS inboxes, to RingEX users, already has more than 5,000 customers. Nearly 40% of them have attached paid AI products. CEB is now available for Microsoft Teams, turning Teams into an informal contact centre with voice, routing, analytics, and SMS inbox capabilities built in.
ACE, the post-call analytics engine, reached more than 5,200 customers. ATB Financial, the largest financial institution in Canada, added RingEX seats and ACE to automate post-call analysis, surface sentiment, and replace manual agent evaluations.
Enterprise Migrations Continue
Several notable migrations reinforced the on-prem-to-cloud pipeline. Coca-Cola UNITED, the third-largest Coca-Cola bottler in the US with 60 locations, is migrating thousands of seats to RingEX. The New York Mets are replacing a decade-old on-premises system with RingEX, RingCX, and Call Queues Booster. Casio consolidated legacy systems onto RingEX and RingCX with ACE Quality Management. A Fortune 500 insurance company is expanding its RingCentral deployment enterprise-wide with tens of thousands of RingEX seats.
On the channel side, Cox Communications has started deploying RingCentral’s AI-powered contact centre to its customer base, with TELUS and Spectrum Business also beginning to bring the AI portfolio to their customers. Shmunis said meaningful GSP-driven AI revenue is “probably not so much for this year, but 2027, 2028.”
Asked whether outcome-based pricing is taking hold in the AI era, Shmunis was sceptical: “I’m not really personally aware of too many people who are actually truly pricing outcomes.”
RingCentral’s model uses monthly subscription tiers with allocated minutes, offering predictability for both customers and the provider. “Everybody knows what they’re consuming. We know our costs. They understand their spend,” Shmunis said.
