Skip to main content

Inside Five9: Voice, AI Agents & Agent Assist Ahead of Dreamforce 2026

Image
Julian's headshot
Julian Tiongson Manager, Product Marketing

Julian Tiongson specializes in product marketing with emphasis on product messaging, positioning, and go-to market initiatives. He is a well-rounded marketer starting his early career from wide range of social media marketing, email marketing, content marketing, and has since transitioned into product marketing in the later stages. This has brought a unique position to his product marketing style with deeper understanding of the full customer life cycle.

A customer shouldn't have to know or care how many systems it takes to solve their problem. What they notice is whether the wait was short, whether the person (or AI agent) they reached already had the context, and whether the issue actually got resolved on the first try.  

In Part 1, we covered how Five9 Fusion for Salesforce gives agents that context in one service console instead of two, and how Five9 supports this ahead of Salesforce's February 2028 Open CTI retirement. This post goes deeper on the capabilities that make both paths worth choosing: the Voice, Voice AI Agents, and AI Agent Assist capabilities doing the work underneath, and the measurable results real organizations are getting from them ahead of Dreamforce 2026. 

Voice: the foundation everything else runs on 

Voice is still the hardest channel to get right in a contact center, and it's the foundation of the Five9 platform. The underlying telephony network is what enables routing, transcription, the AI agents actually runs on top of. This means there's no separate system to reconcile when data needs to move between them. That turns out to matter for AI specifically: agentic voice performs noticeably better on infrastructure designed for it from the start, and it's a big part of why some large enterprises running contact centers on-premises, like Alaska Airlines, have moved to the cloud with Five9

Voice AI Agents: self-service that can actually finish the job 

Five9 Voice AI Agents, with the newest release introduced this summer, represent a real architectural shift away from scripted bots and legacy IVR. A few things set them apart: 

  • Multi-agent orchestration, not a single overworked bot. A coordinating layer breaks a request into steps and hands each one to a specialized agent built for that task, so a longer, multi-step customer journey gets resolved through coordinated teamwork instead of one generalist trying to cover every possible intent. 

  • Conversations that don't feel like talking to a machine. Fast response times, support for multiple languages, and the ability to handle interruptions and background noise cut down on the awkward pauses that give older IVR systems away. 

  • Agents that can act, not just talk. Through tool calling, an AI agent can pull up an account, update a record, or push a transaction through. The kinds of steps that used to require a transfer to a human. 

  • Guardrails for regulated environments. Sensitive data gets stripped out before it's ever passed to the underlying model, and an agent won't hand off a call until it's actually finished what it set out to do. 

  • Warm handoffs to human agents carry full conversational context, so customers never repeat themselves. 

Early results are already showing up in production. Logistics company PODS put Five9 Voice AI Agents into its contact center and came in above its own containment goal, landing at 44% with handle times coming down and call quality staying consistent across a much larger volume of interactions than a pilot typically covers.  

The company is on pace to route more than 100,000 calls through the AI agents by the end of the year. Ruthu Raj, VP of Architecture, IT Infrastructure and Operations at PODS, has credited Five9 with working through the harder technical problems such as background noise, knowing when a caller is actually done talking, keeping the model from making things up that tend to be where voice AI projects stall out. 

AI Agent Assist: help for the agent, not just the customer 

Not every interaction should be automated end to end, and Five9 AI Agent Assist is built for the calls that need a human. It gives agents real-time transcription, live guidance, and automated call summaries as the conversation happens, so agents can stay present with the customer instead of taking notes. That same data feeds supervisor coaching and quality workflows after the call, turning every interaction into a training opportunity instead of a black box. Within Fusion for Salesforce, Five9 AI Agent Assist is one of two AI paths customers can choose, the other being Einstein for Service, powered by the same live data through Five9 TranscriptStream. 

The rest of the contact center toolkit 

Voice and AI agents get the spotlight, but the platform underneath them is broader: 

  • Omnichannel routing across voice, chat, email and SMS, so context follows the customer regardless of channel. 

  • Workforce management and analytics that give supervisors visibility into performance and staffing in real time. 

  • Speech and interaction analytics for quality monitoring, compliance and coaching at scale. 

  • Outbound engagement tools with predictive, progressive and power dialing for proactive, compliant outreach. 

Use cases worth building around 

Across customers, a few patterns show up again and again: 

  1. Self-service automation for high-volume, repeatable requests such as order status, scheduling and account updates all handled end to end by Voice AI Agents. 

  1. AI Agent Assistance for complex or sensitive interactions, where a human agent stays in the seat but gets real-time support. 

  1. Unified CRM and contact center data so every handoff, AI to human, channel to channel carries full context. 

  1. Supervisor coaching powered by call data, turning every interaction into a feedback loop instead of a one-off. 

Talk through your Open CTI migration at Dreamforce 

Reading about agentic voice AI only gets you so far; the difference is obvious once you hear it handle an actual conversation. But if Open CTI's February 2028 retirement is on your radar, the most useful thing you can do at Dreamforce isn't watching a demo; it's talking to a Five9 expert about your specific migration.  

Stop by the Five9 booth at #529 at Moscone Center this September. We'll show you Voice AI Agents and AI Agent Assist working together inside Fusion for Salesforce, and help you map out which path, and which timeline, makes sense for your environment. Bring your use cases and we'll be there to show you rather than tell you. 

That's Smarter CX with every interaction and we'll see you at Moscone Center!

Image
Julian's headshot
Julian Tiongson Manager, Product Marketing

Julian Tiongson specializes in product marketing with emphasis on product messaging, positioning, and go-to market initiatives. He is a well-rounded marketer starting his early career from wide range of social media marketing, email marketing, content marketing, and has since transitioned into product marketing in the later stages. This has brought a unique position to his product marketing style with deeper understanding of the full customer life cycle.

Call 1-800-553-8159 to learn more about Five9