Fin 4 Launch: AI Customer Service Agent for Agentic CX

AI agents are rewriting the rules of customer service. At Pioneer 2026, Fin’s Chief Product Officer Paul Adams unveiled Fin 4, the most significant system-wide upgrade in Fin’s history. The Customer Agent is smarter, Operator is more autonomous, collaboration between AI and human teams has been reimagined, and Fin’s proprietary Apex model family has reached its next generation.

Fin’s goal hasn’t changed: give every customer an outstanding experience in every single interaction. Fin 4 is a decisive move toward that goal and toward truly agentic customer experience (Agentic CX).

Why Now? Three Shifts Reshaping AI Agents

AI capability in 2026 has advanced faster than almost anyone predicted. Twelve months ago, Claude Code was unfamiliar to most people. Today, it produces most of Fin’s code, guided by the Fin team.

Fin sees AI agents changing in three fundamental ways. They are becoming:

More autonomous: they keep working in the background while your attention is elsewhere, and the results are waiting when you return.
More proactive: they act without being asked, picking up on signals and moving forward based on goals or triggers set in advance.
More directive: as agents accumulate more knowledge, they increasingly guide people on what to do next.

These shifts point to a clear destination. For businesses of every size, 90–95% of customer communication will soon require no human involvement. Fin sees this as inevitable; the open question is timing. And the biggest obstacle won’t be the technology. It will be how organizations adapt.

Fin’s Blueprint for Next-Generation Customer Service: Three Layers Working Together

Fin believes the future of customer service rests on three layers.

Layer one: the Customer Agent.

It spans the entire customer lifecycle, understands both business goals and customer context, handles more than 90% of inbound conversations, and reaches out proactively to unblock customers and drive business outcomes.

Layer two: the Operations Agent.

The more powerful a Customer Agent becomes, the more it needs to be managed. Rather than having people click through screen after screen of settings, an Operations Agent can manage and optimize the Customer Agent. That’s exactly why Fin launched Operator in May. Human teams set the strategy and design the experience; Operator carries out the work and keeps improving it.

Layer three: a helpdesk rebuilt for AI.

Ticket-based support systems have defined the past decade, but they weren’t built for this moment. Nobody starting from scratch today would design them that way.

Every update in Fin 4 maps to one of these three layers.

Upgrade 1: Fin Moves From Answering Questions to Driving Customer Success

Fin has already expanded from Service into Sales and Ecommerce. With Fin 4, it now takes on Customer Success, bringing three fundamental changes:

Proactive by default: Fin starts conversations itself, deciding when to step in, what to say, and which channel to use.
Outcome-driven: Fin works toward the business goals you define, across timeframes of days, weeks, or months.
Adaptive to every customer: Fin handles whatever each customer brings up along the way.

In practice, Fin can introduce customers to features they haven’t tried and help them get started; build a tailored onboarding plan for each new customer and guide them through it; re-engage customers whose usage is slipping; spot early churn signals and intervene before it’s too late; and surface growth signals that point to expansion opportunities.

Work that used to require your team to manually review accounts and coordinate outreach one by one now happens automatically.

Upgrade 2: Skills Let Fin Solve Problems That Don’t Have a Script

Fin’s vision is a single Customer Agent handling all customer communication for each business. That means Fin needs to handle every kind of problem, not just the predictable ones.

Last year’s Procedures gave Fin the ability to resolve complex, multi-step requests such as refunds, account changes, and subscription updates, while giving businesses precise control over how their most critical processes run. But plenty of real-world problems have no fixed path. Troubleshooting and issues that require investigation have traditionally landed back on the support team’s desk.

Fin 4 introduces Skills to close that gap. Each Skill combines:

Triggers that tell Fin when to use it;
Tools that let Fin take action inside your systems;
Instructions that set guidelines and guardrails.

Give Fin a job, and it will combine different Skills to work out how to get it done. The more Skills Fin has, the more it can handle.

Upgrade 3: Operator Gets More Proactive, More Autonomous, and Connected to Slack and GitHub

Since its May launch, more than 1,600 customers have used Operator to transform how they run customer experience. Fin 4 gives Operator four new capabilities:

Routines: Operator runs work on a schedule or in response to events, such as sending a team performance report every Monday or flagging an incident with a diagnosis and a suggested fix already attached.
Connections: Operator reads information from tools like Slack and GitHub and acts on it. It can pick up a product announcement in Slack and draft the corresponding changes to Fin, or file a bug report directly in GitHub.
Memory: Operator retains your preferences, from writing style to how you like reports structured, and applies them in future conversations.
Artifacts: Operator’s output, such as a drafted Procedure or an analysis of customer conversations, can be shared with the teammates who need to review or act on it.

At Teachable, the team even had Operator orchestrate an entire payment migration on its own.

Upgrade 4: Human in the Loop and a Second-Generation Copilot Turn Support Teams Into AI Supervisors

When a Customer Agent resolves more than 90% of questions, the role of the human support team has to evolve. Fin 4 brings two major updates to the Intercom helpdesk.

Human in the Loop:

whenever a conversation needs a review, extra information, or an approval, Fin brings in a human teammate to approve, edit, or take over. A confidence score setting lets businesses control when Fin should ask for human sign-off. The support rep’s job shifts from answering questions to supervising Fin.

Second-generation Copilot:

Fin has redesigned and rebuilt Copilot from the ground up as a true agent. By the time a rep opens a conversation, Copilot has already searched every connected tool, summarized the customer’s issue, drafted the best response, and recommended next actions. For example, it can understand the customer’s problem, find the matching open issue in Linear, and prepare a reply the rep simply reviews and sends.

Both features run natively in Intercom and will also integrate with other helpdesk platforms.

Upgrade 5: The Next Generation of Apex Models Raises Resolution, Speed, and Accuracy

None of these capabilities matter unless Fin delivers best-in-class performance. Fin’s proprietary Apex model family already outperforms every commercially available frontier model on real customer service tasks. Fin 4 introduces three new models.

Apex 2.0

Apex 2.0 is Fin’s most capable customer service model to date. It performs better on the core metrics that show a problem was genuinely solved, with more customers explicitly confirming their issue was resolved (a “hard resolution,” one of the toughest metrics in the industry to move). It’s also more firmly grounded in each company’s own content, cutting hallucinations by 16.5%.

Apex Flash 2.0

Apex Flash 2.0 is built for latency-sensitive channels like voice. Built on NVIDIA Nemotron 3 Nano, it’s the fastest model Fin has ever shipped. Fin Voice now runs on Apex Flash 2.0, responding faster, following guidance more closely, and keeping answers concise, which is exactly what phone conversations demand.

Apex for Procedures

Apex for Procedures is an all-new model trained to reason through complex, multi-step queries. In Fin’s testing, it beat Sonnet 5, the strongest frontier model evaluated for this work, across every key metric, delivering 15% more hard resolutions and 37% fewer major hallucinations.

The Future of AI Customer Service Is Already Here

Fin envisions a future where every business has one Customer Agent handling more than 90% of customer communication across the entire customer journey; where Operator manages and optimizes that agent while human teams focus on strategy and experience design; and where agents and people collaborate closely inside a next-generation helpdesk, sometimes with agents guiding humans, sometimes with humans supervising agents.

Fin 4 is a major step toward that future.

👉 Explore Fin 4 today, start your free trial, and experience truly agentic customer service.

DKM Ecosystem is an Intercom APAC Distributor, APAC Gold Partner, and Professional Services Certified Partner, dedicated to helping enterprises and partners adopt, implement, and scale Intercom across the Asia-Pacific market.

DKM Ecosystem delivers Intercom consulting, implementation, integration, and professional services—enabling businesses to build, deploy, and continuously optimize AI-powered customer service systems. Its service capabilities span Fin, AI Agents, workflow automation, omnichannel customer service, system integration, localization, training, and technical enablement—providing end-to-end Intercom support from initial solution design and product deployment through ongoing operational optimization.

🌏 Professional services and localization expertise across APAC and key global markets: Including Mainland China, Singapore, Japan, South Korea, Hong Kong (China), Taiwan (China), Malaysia, the Philippines, Thailand, India, Indonesia, Vietnam, Cambodia, the UAE, and Australia

📡 Omnichannel customer service solutions: Supporting customer interactions across Chat, Email, Voice, WhatsApp, Instagram, Facebook Messenger, SMS, Slack, LINE, WeChat, Lark, DingTalk, and more

🔗 Intercom, Fin, and AI Agent integrations: Seamlessly connect with existing customer service platforms, CRMs, APIs, and core business systems

🤖 AI-powered customer service capabilities: Combine Fin, AI Agents, and workflow automation to elevate service efficiency and automation

🔑 Solutions for 35+ industries: Tailored AI, automation, and real-time analytics capabilities to meet diverse industry-specific customer service needs

📈 Training, technical enablement, and continuous optimization: Help businesses fully realize Intercom’s platform value—and continuously improve customer experience and service efficiency

If you’d like to learn more about Intercom, Fin, AI Agents, and AI-powered customer service solutions, or explore a customized Intercom implementation and integration plan for your business, start your free Intercom trial today, or contact DKM Ecosystem to discuss the right solution for your business.

👉 Learn more about DKM Ecosystem on the official Intercom Partner page:

https://www.intercom.com/solution-partner-program

Fin 4 Is Here: AI Customer Agents Enter the Fully Autonomous Era, With 90% of Conversations Handled Without a Human