Fini

Everyone talks about AI resolution rates. This is what it actually looks like. Sophie is one of our AI agents. In this video, she's handling a real customer question: reading the message, pulling the right information, and resolving it in under two minutes. No escalation. No human in the loop. Just a customer who got their answer and moved on. This is what we mean by resolution, not deflection.

The best AI implementations start with the boring stuff. Kenji Hayward (Senior Director of Customer Support at Front) has seen teams get this wrong. They launch flashy customer-facing bots before fixing internal friction. Then they wonder why adoption stalls. His take: "Solving internal friction first turns AI from a perceived threat into a trusted teammate." Start with the stuff your team hates doing. Password resets. Ticket tagging. Internal lookups. Let them feel the benefit before you put AI in front of customers. Welcome to the Fini Hall of Fame, Kenji 💙

Support leaders don't talk about this, but the real hesitation with AI is accountability. If it breaks, your name is on it. Tamara Wall has been in that seat. She's led teams through COVID, reorgs, and multiple AI rollouts. She once saw engagement swing 55 points during a single change initiative. She's now Head of Support at Common Room, after leading operations at Culture Amp and Datto. Her advice: start small, pilot safely, measure relentlessly, adapt publicly. The risk never disappears. You just get better at managing it. Episode four is live. Link in comments.

Wefunder stopped hiring for support. Response times dropped from 7 hours to 15 minutes anyway. Same team. 2x the volume. Kai Moon from their team: "There's room to breathe now." Full case study: https://lnkd.in/eK4chwXZ

88% of contact centers say they're using AI. Only 25% have actually integrated it into daily operations. The rest? Pilots. Experiments. Partial rollouts. Tools that technically exist but aren't resolving much. If you follow CX conversations online, it feels like everyone is running AI agents and posting 70% resolution rates. Then you call your doctor's office, your insurance company, or your kid's school. The companies posting results on LinkedIn are real. They're also a small fraction of the market. Most businesses are still figuring out where AI fits, what it can handle, and whether they can trust it. For heads of support considering AI: you're not late. The early adopters are ahead, but the industry is still years away from widespread deployment. The gap between "we use AI" and "AI actually resolves our tickets" is where the opportunity is right now.

You can't force trust onto people. But most AI rollouts try to. Sianne Hussey (Customer Director at ProfitPeak) has watched this play out across teams. Her take on what actually works: "Prove the tech works while giving people permission to start small and reclaim their time." Adoption follows small wins, not announcements Welcome to the Fini Hall of Fame, Sianne 💙 Every week we feature CX leaders with a real perspective on AI in support. Want to nominate someone (or yourself)? DM us or drop their name in the comments.

Wefunder is now powered by Fini. Their response time went from 6 hours to 15 minutes. They're handling 2x the volume with the same team. In week two, Fini answered 1,000 emails while the team sent 240. "We can finally shift our focus from reactive to proactive." — Kai Moon, Success Team Lead Full case study coming soon.

Your best support team might have your lowest CSAT score. This ticket was resolved in under 2 minutes. The customer gave it 1 star. Why? Because they're rating their frustration with the billing error, not the quality of the support they received. Your CSAT dashboard doesn't know the difference. This happens constantly. CSAT measures how frustrated customers were before they reached out, not how well your team handled it. A team grinding through complex billing issues will score lower than a team answering simple questions for a stable product. If you're using CSAT as your primary metric for evaluating AI support, you're blending product frustration into a score that's supposed to measure support quality.

Eli Winderbaum has spent 12 years building CX organizations at BetterCloud, Marcus by Goldman Sachs, and now Mirage, where they're handling 10,000 conversations a month. Two years ago, he deployed an AI agent. Today, it resolves 65% of all inbound messages. But the interesting part isn't the automation. It's what happened to his team. His tier one agents didn't disappear. They got promoted. They now handle complex issues, track feature requests, and feed insights directly to product. The AI became tier zero. The humans moved up. His take on what wins in 2026: "The team that is actually listening to their customers will win. If two products are equal in every way, but one gets back to you instantly with a great answer, customers will move." Episode three of the Fini Podcast is live. Link in comments ⬇️

LISA’s Head of Support was spending 30+ hours a week as a human search engine. Three sports federations. Overlapping seasons. The same questions about bookings and schedules, over and over. She tried Zendesk. Too complex. She tried generic AI tools. They hallucinated, mixing up Hockey rules with Tennis regulations. In a regulated sports environment, that killed the project. When she came to us, she was skeptical. Understandably. Two weeks later, half her support workload was gone. What made the difference? The AI actually respects multi-brand context. It knows which federation the question is about. It doesn't guess. And when it doesn't know something, it says so instead of making things up. That's the bar. Not "sounds impressive in a demo." Actually works when the stakes are real. Full case study in comments ⬇️