Pioneer 2026: Fin 4, Operator, and the evolution of human work

This week, more than 600 leaders operating at the frontier of CX joined us in the Regency Ballroom in San Francisco, and thousands more online, for Pioneer 2026.

Since this time last year, AI has advanced massively, with Agents becoming more autonomous, more proactive, and more directive. With these advancements, we can now see a very clear path to a fully agentic customer experience where 90% or more of customer communication will not involve a human.

Our mission is to deliver perfect customer experiences to every customer, every time. Until now, that hasn’t been possible for most businesses, and the average customer experience today is still really poor. But we now have a technology in AI that makes our goal possible.

At Pioneer, we announced Fin 4 – major upgrades to the Fin system that represent a big step toward that mission. Throughout the day we heard from leaders at Anthropic, Salesforce, Clay, Box, AppFolio, Nuuly, Gamma, and Teachable about how they’re transforming their customer experiences with AI.

In the room and in the livestream chat, we heard the same feedback from people – it’s a relief to know we’re all figuring this out.

If you missed the sessions, they’re all available to watch on demand. Here’s a look at what happened throughout the day.

At Pioneer 2025, we announced our vision of Fin as a Customer Agent, capable of handling the entire customer experience. Since then, Fin has taken on the jobs of sales and ecommerce, and, most recently, customer success.

We also launched Operator, the Operations Agent for Fin. We made big upgrades to Voice. We launched Apex, our own models, purpose-built for customer experience.

We changed our company name to Fin and joined Salesforce.

Fin now has 12,000 customers, resolves more than 10 million conversations a month, and averages a 76% resolution rate.

As Eoghan said, “If any of us heard 3 or 4 years ago that you would be able to employ AI that would perfectly, excellently, quickly, to the satisfaction of your customers, handle three-quarters of your service and customer communication needs, we just wouldn't believe it. That's happening today.”

Our Chief Product Officer Paul Adams announced Fin 4, a set of major updates to every part of the Fin system – a more capable Customer Agent, a more powerful Operator, and new ways for Fin to work with your team.

He was joined by our Senior Director of Product Marketing Tom Smith to demo Fin’s new features. You can watch the full keynote and all the demos here.

Here’s what’s new:

• Customer Agent: We believe every business will have one Customer Agent. Ours just got a big upgrade. Fin now drives customer success and, with Skills, figures out how to solve problems without you defining every step.
• Operator: Fully agentic customer experiences need an Agent to run complex operations. Routines and Connections give Operator powerful new ways to do that work for you, proactively and autonomously.
• Copilot and Human in the Loop (HITL): As Fin resolves more, human teams take on a supervisory role. A second-generation Copilot and Human in the Loop give them entirely new ways to work together.
• Fin Apex: The next generation of Fin’s AI models is here. Apex 2.0, which uses NVIDIA Nemotron 3 Super as a base, and Apex Flash 2.0, built on NVIDIA Nemotron 3 Nano, push performance further across resolution rate, speed, and accuracy. Apex for Procedures extends Fin’s specialized model approach to complex, multi-step queries.

Learn more about Fin 4 here.

Our VP of Product Brian Donohue was joined by Gamma’s Alannah Carroll and Teachable’s Kathleen Ross to talk about how they’re running their support operations with Operator.

Kathleen described Teachable’s migration to Stripe for payments, which touched more than 100 help articles. Operator built a readiness plan, helped draft new articles and snippets, and suggested Monitors to catch any potential problems after launch.

A year ago, auditing the 25 most-referenced Fin articles took Kathleen’s team more than a month. “I gave Operator the doc, and this feels extreme to say, but it feels like it was done within an afternoon,” she said.

Alannah calls herself Gamma’s “Fin wrangler.” She got Operator to learn from the best examples of human troubleshooting conversations and turn those into a Procedure for Fin. She’s also set up a daily routine that reviews low-rated conversations overnight, so the fixes are ready for her to approve as soon as she logs on in the morning.

Reflecting on how much Fin and Operator have changed her day-to-day role, Alannah noted that six months ago she was answering support tickets. Now, she’s shipping changes to Fin.

Speaking to the room, Alannah said: “These roles don’t really come with an instruction manual,” but when you run into a hard problem, “Operator usually figures it out, so it helps you feel like you're less alone in this huge sea of Fin things."

You can watch the full conversation here.

Our SVP of Success and Solutions Junan Pang sat down with Natalie Hurst of Nuuly and Jessica Lozoya from AppFolio to dig into what it takes to automate the most complex kinds of work.

Junan opened with a show of hands, asking who in the room had automated FAQs with AI. Almost everyone had. Then, he asked who had automated something more complex like a billing dispute. Only a few raised their hands.

What’s interesting, though, is that technology often isn’t the thing that holds teams back from automating these types of queries. Trust and buy-in are the blockers.

During the discussion, Junan highlighted that Procedures are now involved in over 55% of Nuuly’s Fin outcomes. Natalie shared that her team has automated pausing or canceling a subscription, and refunds – their highest conversation drivers – in a single “mega Procedure.”

Her advice to other teams is: “Don’t wait for everything to be ready. Sometimes you just have to rip that band-aid off.”

Jessica showed how her team got internal buy-in to automate complex work. Their first Procedure for email deliverability won an internal AI hack day, which led to a business case for automating four more processes with projected savings of $317,000 a year.

“We’re not asking for headcount … We’re asking for access. We’re asking for trust,” she said.

Watch the panel discussion here.

Des took to the stage to explore an idea: “we shape our tools and thereafter our tools shape us.”

When Intercom first launched in 2011, businesses didn’t have an easy way to talk to their customers online. Today, around a billion conversations happen between businesses and customers every year.

Des argued that successful tools and businesses do more than earn money; they change the world around them, and then they change us.

Software has always forced a tradeoff between powerful and simple. One piece of software can rarely do both things well. Des shared examples of an airplane cockpit and a tourist kiosk. One requires a hundred hours of training to be able to use but is powerful and efficient, the other is easily accessible to anyone, but is simple and inefficient.

AI removes that tradeoff. With tools like Operator, you can describe what you want to do, and the agent works it out. The agents are going to use the features, and as that happens, our roles and org charts will change. “Most of your job titles probably didn’t exist five years ago,” he said. "And in five years' time, they're all going to be changed again."

As Des put it: “We’re going to have to become the type of people who live in that new world.”

Watch Des’s keynote here.

Our COO Jordan Neill hosted a panel discussion with Anthropic’s Emily Lampert, Salesforce’s Emily Winslow, and Clay’s George Dilthey where we heard about their own transformation journeys with AI.

They discussed their initial expectations of rolling out an AI Agent versus the realities of doing it at scale, as well as the new and interesting challenges support teams face as AI continues to evolve.

Emily Winslow shared that Salesforce leans hard into iteration and experimentation. When you’re deploying AI at their level of scale, you find the edges of your deployment very fast, and as Emily puts it, “often very publicly.” Her focus is getting signals as early as possible to ensure they’re upholding the quality and level of service that Salesforce is known for.

George Dilthey and Emily Lampert both shared the new challenges they’re seeing now that people are using personal agents to reach out to support. AI Agents are persistent and if they’re told not to take no for an answer, they won’t. George and Emily emphasized the importance of adapting your policies to fit this new world and being precise about when conversations should be escalated to a human, and when they should end.

Catch up on the full conversation here.

Our Senior Director of AI Product Management Rati Zvirawa told us that how businesses and customers think about phone support is changing.

For 20 years, businesses have tried to reduce the number of phone calls they get, and customers have learned how to “beat the machine” to get in touch with a person. Rati argued that now, “you don’t have to adapt the machine. The machine now adapts to you.”

On stage, she played a real call from ParkBee, where Fin opened a parking barrier for a driver stuck at the exit in just a few seconds.

Then, our Principal Engineer Gavin Joyce called Fin live on stage to report a lost bank card. In one call, Fin verified his identity, figured out which card he’d lost based on a recent transaction, canceled and replaced it, and also increased his credit limit.

Phone is a tricky channel to automate, but our new voice model, Apex Flash 2.0, has increased resolution rate by 29.5%.

Watch the full session here to see Fin Voice in action.

Our Director of Product Management Christine Zdelar joined our Senior Director of AI Support Ruth O’Brien and Senior Director of Customer Success Deepina Kapila for an in-depth discussion about Fin’s new customer success capabilities, and the blurring line between success and support.

Today, customer success looks like a pyramid divided in three, Dee explained. At the top, you have a small group of high-touch customers, in the middle, the scaled group, who receive a lighter-touch experience, and at the bottom, a digital group, which gets a “broadcast experience.”

The reason for this makeup is because up until this point, it’s been impossible to offer all customers high-touch support with human-only resources. Ruth shared that it’s traditionally been the same in support, with more of a focus on response times – not every customer segment can receive the fastest response. It just simply wasn’t possible with a human-only approach.

With AI, customer support and success teams have the ability to offer those high-touch, VIP experiences to every customer, and make it feel smooth. As Dee and Ruth both said, “don’t ship the org chart.” Customers shouldn’t feel the seams, just the excellent experience you’ve created for them.

Watch the full conversation with Christine, Dee, and Ruth here.

In the final session on the main stage, Eoghan was joined by Aaron Levie, Investor, Entrepreneur, and Co-founder and CEO of Box, to discuss leadership during this period of AI transformation.

Aaron has led Box for 21 years through the cloud and mobile curves. He said that AI is different; that it’s not only a technology change, but “a labor and work transformation.”

He described three ways he looks at work across Box:

• Imagining every role with 10 or 100 times more information.
• Automating the work nobody wants to do.
• Reinvesting the time you save, for example, automating some of customer service to fund customer success.

He made the case that people are still wired for social interaction, and that the goal should be to use AI to “automate out the barriers as much as possible” so people can spend more time on the work that matters most.

Watch Eoghan and Aaron’s full conversation here.