How Fin's own support team reached 83% automation and rebuilt around AI

“Leading the support function for a company building a Customer Agent and AI-first helpdesk is a unique position to be in,” says Declan Ivory, VP of Customer Support at Fin. “Our support team uses the same technology as our customers; the same features, constraints, and expectations. There are few support teams that can genuinely claim to be their own reference customer. We are one of them.”

That position carries real weight. If Fin’s own support team can't demonstrate excellent support, it undermines everything the company is building. When the company shifted its strategic focus in late 2022 to prioritize customer service as its primary use case, our support team had a clear goal: become the best possible example of what AI-first support looks like in practice.

To do that, we couldn’t retrofit AI onto the support operation we had; we had to rebuild around it.

We’re regularly asked by customers to share what worked and what didn’t, and what had to be rebuilt. Here's our team’s journey to success with Fin.

As the company strategy changed in 2022, our support team identified a number of challenges we needed to solve.

1. 1
   New features were launching faster and with greater complexity, driving up conversation volume and the need for technical expertise.
2. 2
   Our existing support policy, which defined our service level objectives (SLOs), was not where it needed to be. First response times exceeded 24 hours in many cases, and coverage was limited to business hours for most customers. Even against SLOs that weren't considered “best in class,” the team was struggling to keep pace.
3. 3
   We needed to open up additional paths to our support team, particularly for website visitors with technical questions and trial customers.

Rising demand, underperforming SLOs, and an expanding scope of who needed support made one thing clear: incremental improvement wasn't going to be enough.

“We made a conscious decision to become Fin’s best reference customer by redesigning our support model entirely around AI,” says Declan.

The key focus areas were:

Building familiarity with AI early

Before Fin was even in beta, the team adopted the Intercom helpdesk’s AI Assist features to give every support rep early exposure to AI-assisted work. This built familiarity and trust across the team and gave the product meaningful feedback from teammates who were using the tools every day.

When Fin was ready for testing, our support team became its first beta customer. Initial rollout was limited to a subset of customers so the impact on experience could be assessed carefully. With no adverse customer reaction and an initial resolution rate above 25%, the team expanded to the majority of customer segments within a matter of weeks.

Creating new disciplines and roles to manage AI

It became clear early that the quality of Fin's outputs was directly tied to the quality of the content it had access to. The team established a dedicated “knowledge manager” role and built knowledge creation into the new product introduction (NPI) process, targeting a 50% resolution rate for every new feature at launch.

Our Senior AI Knowledge Manager, Beth-Ann Sher, came from our existing team and transitioned from the role of help center manager. Like many careers transformed by AI, her role has evolved from administrative to strategic. She used to concentrate on customer-facing, self-serve content, but now she thinks about how this content, and other sources, can be optimized to drive Fin’s performance so we can continue to improve resolution and automation rates. You can learn more about Beth-Ann’s role here.

The team also created a new “conversation designer” role and hired externally for it, recognizing that delivering a good experience through Fin required dedicated expertise.

Our Senior Conversation Designer, Fred Walton, was hired specifically for an AI-first function and focuses on seamless customer journeys with Fin. His role is focused on removing friction and making the handoffs between automation and human support smooth, always keeping customer satisfaction top of mind. You can learn more about him and his day-to-day work here.

Embedding a continuous improvement culture

“We built a practice of ongoing content improvement across the team, allocating dedicated time outside of the inbox for content development and review,” says Declan. “The goal was to make Fin better every week, not just at launch.”

As Fin continued to evolve, we adopted new features as early as possible:

• Guidance rules to deliver more personalized experiences.
• Procedures to handle complex, multi-touch queries.
• Insights to give the team full visibility into Fin’s performance and customer experience quality. The CX Score has been particularly valuable: it gives an objective read on every customer interaction, opening up new ways to identify and address poor experiences before they compound.

Redesigning the organization

As Fin's role expanded, we needed to evolve the shape of our support organization. We established three distinct functions:

1. 1
   AI Support, led by Ruth O’Brien, focused on training, optimizing, and expanding Fin.
2. 2
   Human Support, led by Bobby Stapleton, handling the complex queries that require deep expertise and judgment.
3. 3
   Support Operations & Optimization, led by Caroline Glackin, responsible for workforce management, quality, enablement, and continuous improvement.

Getting the organizational structure right is important, because without ownership, AI fails. Each function has a distinct mandate and area of ownership, but is working towards the shared goals of delivering excellent customer experiences and continuously improving what Fin can do.

What changed for the human team

As Fin took on more support volume, what reached a human support rep changed completely.

"Every conversation that reaches a human now is genuinely complex," says Bobby Stapleton, Senior Director of Human Support. "Cross-functional, technically demanding, requiring judgment that only comes from deep product knowledge."

This required a complete reset of roles and expectations. Support specialists became technical support specialists and technical support engineers, and where the team spent time changed from answering questions to areas like content management, optimization, and consultative support. The team used the Intercom helpdesk’s Workflows feature to build skills-based routing, ensuring that when a customer needed a human, the handoff went to someone with the right expertise.

Support used to scale by adding people; now it scales by improving the Agent and freeing up our humans to focus where they add the most value. As Bobby says, "We're not just responding to problems. We're proactively reaching out and providing consultative support to customers on how to get more from Fin. We're contributing to adoption, retention, and revenue."

“Our new operating model has also made support a proving ground for talent,” explains Declan. “In the past year, more than 40 people have moved from the support team into roles across the company in product, engineering, sales, and beyond.”

When AI removes the repetitive work, support becomes a launchpad for career growth.

In the time since our team introduced Fin, our overall support demand has increased by nearly 300%. The team has absorbed that growth without a corresponding increase in headcount, all while improving the customer experience we deliver.

“Fin now automates 83% of our total support volume. We’ve built a new support policy around ambitious response targets, 24/7 coverage, and outbound phone support, and have created new access paths for trial customers and website visitors who previously had no route to the team,” says Declan. “Based on current resolution and automation rates, we’ve avoided adding over 100 headcount to handle the volume Fin now manages, an annualized cost avoidance in the region of $7.5M–$9M.”

One initiative, led by Franka Martinovic, Director of Customer Support, created a consultative team that focused on proactive engagement with self-serve customers – those without a dedicated sales or success touchpoint. Their goal was to give this group the ability to talk directly with members of our team who had first-hand experience with AI transformation and help them see the value they could get from Fin.

“Over a six month period, we tracked feature adoption, Fin usage, and expansion revenue across accounts we engaged with and accounts we reached out to, but didn’t hear back from.” explains Franka. “Among the accounts we engaged with, the result was an almost 2x increase in both usage and expansion.”

As Franka adds, “In an AI-first world, where Fin handles all of the transactional work, this kind of work becomes even more important. Because the question for support leaders is no longer ‘how do we handle more tickets?’ but rather, ‘how do we use support to grow the business?’”

Looking ahead, our team’s next priorities are:

• Expanding the knowledge and content available to Fin to push resolution rates further.
• Extending Fin into more customer interactions and journeys.
• Using CX Score and Insights to continuously assess and improve the experience for every customer interaction.
• Using Fin’s advanced features, like Guidance and Procedures, to help us automate more work.
• Investing in the human team's ability to take on increasingly complex, consultative work with confidence.

“We're developing a plan to get us to 95% automation," says Ruth O’Brien, who leads the AI Support team. "This isn't a vanity metric; every percentage point reflects a better experience for our customers."

Beyond that, we’re working with other teams internally like sales, success, and billing operations to apply our learnings with Fin to other stages of the customer journey.

"In our latest research, we found that more than half of organizations now plan to scale AI to other parts of the business in 2026, and the majority cite support’s success as the reason,” says Declan. “The operating model, responsibilities, and continuous improvement discipline we’ve implemented in support – all of it is transferable. What support teams build becomes the foundation for how the broader organization can deploy AI across the customer lifecycle.”

You can learn how to launch and scale an Agent for customer service in The Blueprint.