1
At a glance

Company

Anthropic is an AI research company that focuses on building safe, reliable artificial intelligence that’s aligned with human values.

https://www.anthropic.com/

Headquarters

San Francisco, California, United States


Industry

Artificial Intelligence (Al)


Key features used

Fin AI Agent
Help Center

Pioneering AI safety and research company Anthropic is working to build safe, reliable artificial intelligence that’s aligned with human values. When it came to choosing an AI agent to power its customer service, Anthropic wanted to partner with a company that was developing AI technology in line with that ethos – which led them to Intercom.

We are committed to delivering exceptional user experiences, as we know this is a key piece of ensuring transformative AI goes well for humanity.

Customer service plays a fundamental role in Anthropic’s mission of ensuring transformative AI helps people and society flourish. It enables the team to gather critical feedback that shapes product development and sets customers up for success using artificial intelligence. "We are committed to delivering exceptional user experiences, as we know this is a key piece of ensuring transformative AI goes well for humanity,” says Isabel Larrow, Product Support Operations at Anthropic.

After careful consideration, the Anthropic team chose to implement Intercom’s Fin AI Agent to help them achieve this. Since implementing Fin just over a month ago, they have achieved a resolution rate of 50.8% and automatically resolved tens of thousands of customer queries, enabling them to scale personalized support – and help more and more people safely make the leap into an AI-first world. “We’re only a month into this journey, and honestly, it’s hard for us to imagine how things happened before,” says Emily Lampert, Head of Product Support at Anthropic.

The results Anthropic has seen in just a few short weeks speak volumes, but some burning questions remain: what challenges were the team trying to solve with an AI agent; what tipped the scale in favor of buying one instead of building their own – and specifically, what tipped it in favor of Fin?

Let’s find out.

2
Tackling the problems to be solved

When Anthropic decided to more deeply invest in an AI-enabled customer service solution, there were two core challenges that needed to be addressed:

1. Tailoring support to meet a wide range of customer needs at scale

Anthropic has a large customer base that spans everyday consumers using free Claude accounts to large organizations using its SaaS and API products. This diverse set of plan tiers and customer types brings with it a lot of unique needs that need to be met – and support queries of varying levels of complexity.

The team knew that without AI, their ability to provide highly tailored customer service at that level of scale would be extremely difficult, which prompted their search for the right AI agent. “We want to make safe AI available to everyone, so we try to support our customers in different ways,” says Isabel. “The challenge we faced was that we needed to find more scalable ways of supporting them. And AI really was a critical part of the solution.”

2. Managing sharp spikes in support volume

AI technology is growing more sophisticated by the day. As a pioneer in the space, Anthropic’s platform and products are constantly evolving, with new features and models being added all the time.

We knew we needed to smooth out that volume curve and make sure that our team could weather those peaks and remain focused on the most complex issues.

The domino effect is that Anthropic’s support volumes regularly spike, and without the right AI-powered tools to manage these surges, the support team could easily become overwhelmed. “We knew we needed to smooth out that volume curve and make sure that our team could weather those peaks and remain focused on the most complex issues,” says Emily.

3
To build or to buy: choosing the right AI solution

As a research lab, Anthropic could have easily chosen to build an in-house AI agent to power its customer service. But the team saw a greater opportunity to partner with a company with deep expertise in AI-first customer service so they could remain laser focused on their goal of building safe, reliable, and transformative AI for humanity.

We’re an AI company, so if someone contacting us has a poor experience with our implementation of AI, we’re failing to represent ourselves in the right way.

As Emily and Isabel share, there were a number of factors that led to the decision to use Intercom’s Fin over building their own solution:

  • Trust: “We knew we wanted to leverage Claude, our own AI, because we were sure we could trust its quality and safety. Once Intercom started using Claude to power Fin, it was a major unlock,” says Emily.

  • Expertise: “We recognized that Intercom is a true expert in the field and we wanted to partner with a company that’s obsessed about customer service,” Emily shares. “We saw that the Intercom team is constantly working to make Fin more accurate and reliable, which made our decision pretty easy.”

  • Speed: “We knew that we could unlock the power of Fin in a week or less, and that’s what we needed,” Emily adds. "We experience big waves of support volume and needed an AI agent that could reliably handle those influxes – we had confidence that Fin could do that.”

  • Shared commitment to quality: “When we started looking into adding AI to our support experience, we had a high bar for customer experience,” says Isabel. “That’s our brand – we’re an AI company, so if someone contacting us has a poor experience with our implementation of AI, we’re failing to represent ourselves in the right way. One of the most important considerations for us was that brand piece – finding a company we believed shared our quality bar and that could execute AI-first customer service really well.”

  • Innovation: “I attended Pioneer, Intercom’s customer conference, this year and was impressed with the AI-first approach, useful and relevant new features, and range of ways that customers are succeeding with Fin,” Emily notes. “So many other players in the field have gone stagnant – they haven’t moved forward. Intercom is one of the main companies taking an innovative stance in the space.”

If you’re debating whether to build your own AI solution or buy one, as a fast-growing company in a complex space, my advice would be to buy – and specifically, buy Fin.

The “build vs buy” question is one of the biggest considerations when implementing any new technology. When asked what advice they’d give to other support leaders facing the same conundrum, Isabel shared, “If you’re debating whether to build your own AI solution or buy one, as a fast-growing company in a complex space, my advice would be to buy – and specifically, buy Fin. Intercom combines safe, reliable, and cutting-edge AI built by Anthropic with decades of customer support expertise and an AI-first approach, which means you can deploy a world-class solution in days or weeks instead of months, with the confidence that your brand and users are in good hands.”

case-studies-anthropic-to-build-or-to-buy-choosing-the-right-ai-solutioni-fg
4
The journey from setup to success and beyond

Phase 1: Setup and testing

For the Anthropic team, setting up Fin was fast and straightforward. Using their own AI, Claude, to prepare their content and generate snippets from their knowledge base and common support questions to train Fin, the team was able to get up and running in no time.

“In under a week, Fin was live and answering questions,” said Isabel. “It was straightforward to implement, and we were able to use Claude to generate new snippets and restructure macros, ensuring Fin delivered accurate, thoughtful responses from day one. The blend of our two technologies really helped me to get things set up quickly, in a really efficient way.”

Leveraging Intercom’s help center and Fin features are now key cornerstones for how we support free users at scale.

Once Fin had hit the ground running, the team started developing a “free user support strategy” to maximize its impact. Their goal was to leverage Fin to handle a bulk of the queries from free users – which are often repetitive in nature – to take some of the weight off their human agents and increase capacity to focus on more complex queries. “Our free user support strategy is an AI-first approach,” says Isabel. “It involves leveraging our self-serve content so our free users can find answers to more transactional questions quickly, freeing up our team to spend extra time with our enterprise and power user customers that often have more complex questions that need more investigation."

Adding to this, Emily shares, “Fin was a major unlock when it came to being able to implement this free user strategy. Leveraging Intercom’s help center and Fin features are now key cornerstones for how we support free users at scale.”

Phase 2: Early signs of success

Within just a few weeks of using Fin, the wins quickly started to roll in:

  • 96% Fin involvement rate: Fin is involved in almost all customer conversations across both email and the Messenger.

  • 50.8% resolution rate: More than half of all customer queries are now resolved by AI, up from 36% at launch. The team is continuing to optimize Fin’s performance, with a goal to drive that resolution rate up even higher in the near future.

  • Tens of thousands of queries resolved: In the first month alone, Fin has been able to handle an enormous chunk of Anthropic’s support volume, taking pressure off the support team and freeing them up to focus on complex and strategic queries.

  • Faster response times: Fin’s instant answers have led to overall decreases in response times, a key metric for measuring the quality and success of support.

  • Around-the-clock availability: With a primarily US-based support team, being able to offer 24/7 support is a huge benefit to Anthropic customers around the world.

“The time savings have been huge – in the first month, we'd already saved over 1,700 hours of our team's time," says Emily. "Without Fin, it would have taken us months to work through that volume. And our users appreciate the instant, accurate answers.”

Phase 3: Analysis and optimization

As Anthropic knows all too well, AI is a technology that constantly improves and requires ongoing optimization – so the team’s success with Fin didn’t begin and end with the early results they saw.

Working closely with Intercom, Anthropic made tweaks to improve Fin’s performance using built-in features like its “unanswered questions” report to pinpoint areas for improvement and ensure Fin was always delivering up-to-date, relevant answers. “Intercom’s team was instrumental in guiding us through these optimizations,” says Isabel. “Their support helped us achieve quick wins and set us up for long-term success.”

In an effort to multiply these quick wins, Anthropic set up an internal “hackathon” to get the whole team involved in fine-tuning Fin – and collectively contributing to its success.

Deep dive: Anthropic’s Fin hackathon

“We wanted to refine Fin’s content, so as part of the hackathon, our specialists reviewed and reorganized snippets, updated help center articles, and identified content gaps," says Isabel. "That led to us doubling the number of available snippets, and increasing Fin’s resolution rate by 5.5% in just one week. This small investment of time had an immediate impact, showing just how important well-maintained content is to AI performance.”

case-studies-anthropic-the-journey-from-setup-to-success-and-beyondi-fg
5
Paving the way forward

Anthropic and Intercom have more than one thing in common, but what stands out is a shared commitment to innovation and exploration – with people at the core.

image-mike-and-des-quotes

Looking to the future, Anthropic sees huge potential in what’s to come for Fin. The team is currently testing Fin Actions – a new capability that enables Fin to take actions on behalf of customers by accessing information from their data sources and systems – to make their refund process more efficient.

“The ability to automate refunds would be a gamechanger for us. It could prevent thousands queries from reaching our team each month,” explains Emily. “These refunds are currently a tedious task and not the best use of a human’s time. I would rather have our human support specialists use their expertise to provide a more tailored, context-aware answer where it’s most needed.”

The strongest validation of Intercom's approach comes from Anthropic choosing Fin as their customer service AI agent.

Together, Anthropic and Intercom are proving that scalable, human-quality AI support isn’t just a possibility – it’s a reality, and it’s happening right now. "The strongest validation of Intercom's approach comes from Anthropic choosing Fin as their customer service AI agent,” says Emily. “We aspire to be an AI-first support team, and Intercom is an ideal partner on this journey.”