Good AI, like good bourbon and good barbecue, comes down to the recipe. That was the theme of Replicant's latest virtual tasting event, where our marketing lead Veronica French welcomed our guests for an hour of whiskey smashes, homemade BBQ sauce, and a candid conversation about scaling voice AI in a regulated industry.

Koen Van Duyse, Director of Solutions Engineering at Replicant, sat down with Loren Andersen, Senior Director and Head of Customer Operations at Avant, for a fireside chat on how the fintech lender scaled AI-driven customer service.

The numbers behind Avant's automation

• ~70% call containment, up from 40% with the legacy IVR
• 45% of calls fully resolved by the AI agent, with another 10–15% partially resolved before handoff
• 200,000+ CSAT responses at a 4.7 out of 5 average since launch
• 35–40% efficiency gain in active accounts handled per staff member

What was covered

• What pushed Avant to explore automation? Avant runs a fully outsourced, offshore contact center split between Manila and Montego Bay. Rising account volume, ramp-up time for offshore agents, and cost pressure all pointed toward automating routine call drivers beyond what a legacy IVR could handle.
• Containment or resolution — which matters more? Both. Avant's DTMF IVR contained about 40% of calls. Now, with Replicant’s AI agent, that number is approaching 70%. The improved resolution rate rounds out the full picture of what’s actually happening. With 45-60% resolution, including fully resolved calls, plus those transferred-post-resolution or for out-of-scope requests.
• What did the data reveal? Before Conversation Intelligence, Avant relied on CRM notes and workflow logs to guess why customers were calling them. Now they can see call drivers, intent in detail, and business gaps, including a surprising discovery that heavy call volume about cash advance limits was missing from the app and website.
• What's the business impact? Since launch, Avant has logged over 200,000 CSAT responses averaging 4.7 out of 5, along with meaningful efficiency gains of up to 35-40% per staff member.
• Where's the opportunity ahead? Outbound use cases for loan origination and early-stage collections, plus a chat channel launching in Q4.
• What's the advice for other regulated teams? Stay deterministic if you are in a regulated industry. Avant deliberately scoped flows tightly to manage EFTA and TCPA compliance risk rather than building toward a fully open-ended agentic system. Avant recommended leaning on Replicant’s white-glove service, Replicare. It was a game-changer for their scalability, letting them build out 10 workflows while continuously iterating to improve containment and resolution rates. 

From insights and intelligence to an automation roadmap

Koen closed the segment by walking the audience through the path many Replicant customers follow: start with Conversation Intelligence to get full visibility into every interaction, identify the highest-volume, highest-effort call drivers, and get a roadmap of workflows that could be automated. Once you have that, you can decide which workflows to automate. This is how Replicant works with customers and prospects: by giving a callable AI agent in one hour, and letting customers/prospects prove its value by getting the insights and intelligence they did not have, so they can build AI agents for strategic workflows 

Avant's story is a clear example of that path: start with data, understand why people contact your organization, and automate high-volume, repetitive workflows.

See what Conversation Intelligence could reveal about your own contact center. Try the Voice AI Sprint → Get a working AI agent trained on your own conversation data. Connect to find out more. 

‍