We Brought 200 CX Leaders Together at Crescendo Live. Here's What We Learned

‍

1. You can't bolt AI onto broken workflows and expect transformation.

Our CRO, Michelle Donnelly, asked the room a question most CX leaders have lived: How many times have you found a solution, done the due diligence, priced it out, got it implemented, and the results fell short?

That keeps happening because companies are bolting AI chatbots onto old ticketing systems. Putting AI on top of something that doesn't work very well will not magically fix it. It needs to be AI-native. The steam engine didn't make horses faster. And the better AI gets, the more we need great humans, not fewer. AI handles the routine questions so human agents can do the harder, more complex, more empathetic work.

‍

2. AI adoption is a cultural transformation as much as a technical one; pretending otherwise is a recipe for disaster.

Karan Singh, COO at Headspace, was candid. Their “State of Mind” report found AI anxiety tops the list of workplace concerns: 92% of employees surveyed reported chronic stress. His approach: name the unspoken, and block time every week to actually build — hands on keyboard, not theoretical. The best way to get over the hurdle is to start.

Jessica DiLullo Herrin pushed further: knowing the earth is round and successfully crossing an unknown ocean are two different problems. Organizations that stop drawing the old org chart and stop managing the old metrics are the ones who'll reach places competitors didn't know were on the map.

‍‍

3. Customer service is now a growth engine.

We ran a study that showed service sells better than sales. Lovepop saw a 3-4x improvement in conversion from its support chats, and 7-hour email response times dropped to 18 seconds in four weeks. Their AI assistant Joy is driving 25% of revenue. They moved chat from the back of the site to the front, engagement doubled, and AI handled all of it.

Before Crescendo, I spent a decade at one of the largest CX technology companies, I thought I knew everything about CX. Then, I was asked to run a 600-person customer service organization and realized I knew nothing about running one. So much time gets spent surviving – managing escalations, dealing with the daily grind – that there's no room for transformation. But with AI, customer care doesn’t have to remain a cost center; it’s now a storefront and intelligence engine.

‍

4. The best teams know exactly where humans need to show up and they don't compromise.

SQUIRE's barber community doesn't want tickets or knowledge bases. They're relational – they want conversation, the same way they connect with clients in the chair. So they built CX around ongoing dialogue, not closed cases. 

Their Chief Customer Officer, James Green, made a point that stuck with me: he started out focused on deflection, minimizing cases to human agents, but realized that didn't lead to better customer outcomes. The shift asked a different question: where is the customer going to get the best experience? Sometimes that's AI. Sometimes it's a human. But you can't answer that if your only goal is to deflect. 

Good Eggs treats every food delivery issue as personal, because it is. They built their AI around that reality and saw first response times drop from a full day to under 15 minutes, with net promoter scores that are best-in-class for online grocery. 

Morgan Donaldson from Cntrl+ made the point that any emotional cue — shame, doubt, confusion — is an automatic escalation to a live agent. Empathy isn't optional; it's the cornerstone of trust.

Polywood's Oliver Hart is creating entirely new roles — design support, proactive outreach — things that weren't possible when the team was buried in reactive tickets. The shared philosophy is that AI should reduce friction, not reduce humanity.

‍

5. The companies pulling ahead are learning by doing.

Karan described the stages honestly: experimentation, operationalization, integration, and reinvention. You can't skip to the end. 

ServiceNow's Anshuman Didwania reinforced this – his team 10x'd agentic consumption in six weeks, starting from one team sharing what worked.

Our own story mirrors it. We won Best of Show at Enterprise Connect – competing against teams 50x our size who've been building for 10-20 years. Our Chief Product Officer, Tod Famous, demonstrated how agentic coding can collapse what used to take 30 days into 30 minutes. Not just faster but fundamentally different in what becomes possible.

That spirit – try things, share what you learn, keep moving – was the throughline of the entire day. We closed by recognizing customers whose work reflected exactly that: Lovepop, Nothing Bundt Cakes, Headspace, Grindr, and SQUIRE. Nobody pretended the challenges weren't real, but the energy was unmistakable.

If you're rethinking how your CX operation works, or curious what AI-native looks like in practice, I'm always happy to compare notes. See how our customers are making this real here:  https://crescendo.ai/case-studies

‍

‍