Webinar Recap: How Insurely Introduced Voice Agents To Their Contact Center

Most insurance customers dread calling their provider. The experience is slow, impersonal, and full of friction. Insurely set out to change that - and they did it with voice AI.

In a recent CCW Digital webinar, ElevenLabs GTM leads Jack Piunti and Josh Pimentel sat down with Insurely CTO Colby Garland to walk through how Insurely deployed ElevenLabs voice agents to automate their contact center, handle outbound sales calls, and deliver real human-like support at scale.

Why Voice AI Matters in Insurance

Voice is not a legacy channel. According to CCW Digital research, 40% of contact center leaders say voice is a top priority when sourcing new technology. Customers turn to the phone when the stakes are high. In insurance, that is almost every call.

The problem is that traditional IVR systems and basic chatbots can only process what a customer says. They cannot detect that someone is panicked about a leaking roof or anxious about their coverage lapsing.

So Insurely turned to ElevenLabs proprietary speech-to-text model to annotate calls with emotional context — frustrated, relieved, confused — so the agent responds to how a customer feels, not just the words in the transcript.

Behind the Agent: Insurely

Insurely is an online-only insurance provider for Canadians. No physical offices. No in-person agents. Their entire customer relationship happens through digital and voice channels.

When Colby Garland joined as CTO, the team faced a familiar problem. Customers were starting online quotes for rental home insurance and dropping off mid-form. The forms were long, multi-step, and confusing for first-time renters. Human agents were calling these customers back to walk them through the process. It was manual, repetitive, and hard to scale.

Building the Agent

Insurely built their first ElevenLabs agent, Emily, to handle exactly that.

Emily is an outbound sales agent that calls customers who have started but not completed a rental quote. She walks them through the form step by step, answers insurance questions in plain language, and encourages customers through the process. Colby noted that customers were completing quotes they otherwise would have abandoned — not just because Emily answered questions, but because she was genuinely encouraging throughout.

From there, Insurely expanded into customer care. The agents now operate 24 hours a day, seven days a week.

The business impact is measurable. Support case volume is down. Sales numbers are up. Human agents are having better conversations because routine questions are being handled by AI. Insurely is maintaining headcount while growing throughput. And the quality of every customer interaction has improved, not declined.

How The Handoff Works

One of the most technically detailed parts of the conversation was around call routing and handoff logic.

Insurely runs office hours Monday through Friday, 8 AM to 5 PM. When a customer needs to speak with a human agent, the AI checks whether the office is open and whether a human is available to take the call.

If the office is open, the call is transferred via Twilio. The full call transcript is pushed to the internal system in real time so the human agent has complete context before they say hello. No re-explaining. No context loss.

If the office is closed, the AI offers to book a callback meeting for when the team is back. That booking happens inside HubSpot, mid-conversation, without any manual intervention. Meeting booked, transcript logged, notes summarized — all before the call ends.

The customer gets a seamless experience from AI to human, and Insurely never has to touch it.

Best Practices From Insurely

1. Keep agents narrowly focused. Give each agent one job. Do not build one large agent that tries to handle everything. ElevenLabs makes it easy to transfer between agents as a conversation moves through different stages. Narrow agents consistently outperform broad ones.
2. Invest in your knowledge base. Good documentation helps human agents onboard faster and helps AI agents perform better. If your knowledge base is outdated or vague, your agent will reflect that. Clear, specific, current information is the foundation of a reliable agent.
3. Solve for data before conversations. Insurely's team discovered that asking callers to spell out their email address mid-call created friction. The fix was integrating with their CRM early in the call flow so the agent could pull existing contact data and skip redundant questions.
4. Handle time zones explicitly. Date and time errors were among the earliest bugs Insurely encountered. Setting time zone context before any scheduling action is now a standard step. ElevenLabs has since built native time zone configuration directly into the platform.
5. Define success criteria per conversation type. ElevenLabs included a built-in analysis section where teams can configure custom success criteria and conversation categories. Colby also built external tooling via the API to analyze conversations from multiple angles. Structured measurement leads to structured improvement.
   

Watch the full session

Watch the full session here.