Why Anthropic Chose Fin Over Building Its Own AI Customer Support Solution — 50.8% Resolution Rate in One Month

Why Would a Top-Tier AI Company “Buy” an AI Customer Support Agent?

Anthropic is the developer of Claude—and one of the world’s most closely watched AI safety and research companies. Logically, a team like this has all the capability to build its own AI customer support agent.

Yet their final decision was: to procure Fin—not build in-house.

The results speak for themselves:

96% of customer conversations involve Fin
Within just over a month of launch, AI resolution rate reached 50.8% (up from 36% at launch)
Over 1,700 hours of team time saved in the first month
Tens of thousands of customer issues resolved automatically
True 24/7 global support

So—how did Anthropic weigh “build vs. buy”? And how did they achieve measurable impact with AI customer support in just a few weeks?

Two Core Support Challenges: Complex User Segmentation & Exploding Query Volume

For Anthropic, customer support isn’t just about answering questions—it’s a critical channel for gathering product feedback and helping users truly succeed with AI. But as the business scaled, the team faced two fundamental challenges.

Challenge One: Diverse user types with vastly different needs.

Anthropic serves both individual users on the free tier of Claude and large organizations using its API and enterprise products. Different plans and customer segments generate highly varied, complex inquiries—making it nearly impossible to deliver personalized service at scale using human agents alone.

Challenge Two: Frequent releases of new features and models trigger periodic surges in support volume.

As a pioneer in AI, Anthropic ships updates rapidly—and every major release triggers a spike in support requests. Emily Lampert, Head of Product Support, explained that the team needed to “flatten this traffic curve”: enabling support staff to handle peak loads while reserving their bandwidth for the most complex cases.

Build or Buy? Anthropic’s 5 Reasons for Choosing Fin

As a research lab, building an AI support agent in-house wouldn’t have been technically difficult for Anthropic. But the team preferred to stay laser-focused on its mission—building safe, reliable AI—while entrusting customer support AI to a true domain expert.

1. Trust: Fin is powered by Claude

Anthropic has deep confidence in the quality and safety of its own models. The fact that Fin uses Claude as its foundational model was a decisive factor in their evaluation.

2. Expertise: A company obsessed with customer support

The Fin team continuously improves answer accuracy and reliability. That singular focus made vendor selection straightforward.

3. Speed: Live in under a week

Facing wave after wave of support spikes, Anthropic needed a solution that could deploy quickly and scale reliably—and Fin delivered, going live in a week or less.

4. Brand alignment: Shared standards for experience quality

Isabel Larrow, Head of AI Support, noted: “As an AI company, if users encounter a poor AI experience in support, we’re failing to represent our brand correctly.” Finding a partner whose quality bar matched theirs was non-negotiable.

5. Innovation: A product that keeps evolving

After attending Fin’s annual customer conference, Pioneer, Emily observed that while many competitors stagnate, Fin continues shipping practical, AI-first features.

For other support leaders wrestling with “build vs. buy,” Isabel’s advice is direct: “If you’re a fast-growing company operating in a complex domain, my recommendation is to buy—and specifically, to buy Fin.”

In her view, Fin combines Anthropic’s secure, cutting-edge AI capabilities with deep domain expertise in customer support—enabling enterprises to deploy world-class solutions in days or weeks, not months.

From Launch to Impact: Anthropic’s Three-Step Pathway

Step One: Train Fin with Claude—go live in under a week

Deployment was fast and frictionless. The team used Claude to structure knowledge base content, generate knowledge snippets from common questions, and rebuild support response templates—so Fin delivered accurate, thoughtful answers from Day One. Within a week, Fin was officially live.

They then launched an AI-centric **“Free User Support Strategy”**: Fin and the Help Center now handle the high volume of repetitive, transactional queries from free-tier users—freeing up human agents to focus on complex needs from enterprise and power users. Today, Intercom Help Center and Fin form the backbone of Anthropic’s scalable support for free users.

Step Two: Impact across the board—in just weeks

96% participation rate: Fin engages in virtually every conversation across email and Messenger channels
50.8% resolution rate: Over half of all issues are resolved directly by AI—and the rate continues to climb
Tens of thousands of issues auto-resolved: Significantly reducing pressure on support teams
Faster response times: Instant replies drive down overall response latency
24/7 coverage: With the core team based in the U.S., Fin delivers round-the-clock support to users worldwide

Emily noted that the 1,700+ hours saved in Month One would have taken the team months to absorb without Fin.

Step Three: Continuous optimization—One internal hackathon boosted resolution by 5.5% in a week

AI support isn’t “set and forget.” Anthropic and Fin collaborate closely—using built-in analytics dashboards to identify weak spots and iteratively improve answer quality.

To amplify impact, Anthropic ran an internal **“Fin Hackathon”**: Support experts reviewed and restructured knowledge snippets, updated Help Center articles, and identified content gaps. As a result, the number of usable knowledge snippets doubled, and Fin’s resolution rate rose by 5.5% in one week.

This case makes one thing clear: The quality of your knowledge content directly determines your AI agent’s performance.

Next Step: Empowering AI Agents to Take Action

Anthropic is now testing Fin’s next-generation capabilities: connecting to business data and systems so the AI can act on behalf of customers—for example, processing refunds automatically. Emily believes automated refunds could deflect thousands of tickets per month—freeing human experts to apply their judgment where it matters most: on nuanced, context-rich, highly personalized interactions.

What Both Sides Say: A Partnership Built on “Experimentation & Evaluation”

The relationship between Anthropic and Fin goes far beyond “customer and vendor.” One side builds frontier foundation models; the other brings deep expertise in AI-powered customer support. Together, through continuous experimentation and validation, they’re advancing the state of AI support. As demonstrated by the hackathon and ongoing optimization, this collaborative model itself is key to real-world AI impact.

“Our partnership with Fin is exactly the kind I love—it’s grounded in constant experimentation and evaluation.”
— Mike Krieger, Co-founder of Instagram, former Chief Product Officer at Anthropic

“Our partnership with Anthropic keeps us at the forefront of AI—enabling us to explore the future, dive deep into cutting-edge technology, and deliver stable, reliable, and truly impressive results for our customers.”
— Des Traynor, Co-Founder & Chief Strategy Officer, Fin

Conclusion: The Strongest Endorsement Comes from an AI Company Itself

When the company behind Claude entrusts its customer service to Fin, that’s the most powerful validation of Fin’s capabilities. Anthropic and Fin are jointly proving one thing: scalable, human-level AI customer service isn’t a vision for the future—it’s here today.

Frequently Asked Questions (FAQ)

Q: Should enterprises build or buy AI customer service?
For fast-growing, operationally complex businesses, purchasing a mature AI customer service agent typically enables faster time-to-value and lets teams focus on core business priorities. After choosing Fin, Anthropic went live in under one week.

Q: Which AI models does Fin use?
Fin uses Anthropic’s Claude as one of its foundational models—balancing response quality with safety. The two companies maintain close collaboration across both model and product layers, continuously experimenting and evaluating performance.

Q: How can AI customer service resolution rates be improved?
The key lies in the quality—and continuous optimization—of knowledge content. Anthropic boosted its resolution rate by 5.5% in just one week through an internal hackathon dedicated to refining its knowledge base.

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

Why Anthropic Chose Fin Over Building Its Own AI Customer Support Solution — 50.8% Resolution Rate in One Month, Saving 1,700 Hours