AI First Customer Service – Paul Adams blog hero

There’s no going back – the new era of AI-first Customer Service has arrived

The era of AI-first Customer Service is here, and it will lead to happier customers and more strategic customer service teams.

It’s already clear that AI will reshape society, just as major technology breakthroughs have done in the past. Technology doesn’t go backwards. The car created shopping malls, suburbia, and McDonald’s. The internet created instant communication, real-time shared events, on-demand everything, for anyone, anywhere on earth. We’re not going back to riding horses, nor printing out photos to post to our family across the world. AI is as big an invention as the internet, and possibly as big a change as the industrial revolution. We’re only 18 months into a 10+ year cycle.

“Almost all customer questions, in all customer channels, will be excellently answered by an AI Agent”

Customer service is one of the earliest areas to be fundamentally changed, because AI happens to be excellent at the things customer service people do today. This transformation is already happening at a faster pace than anyone predicted. Based on results we’re seeing from our customers (some already have up to 80% of customer questions being accurately answered by AI), we can now confidently predict that in the very near future, almost all customer questions, in all customer channels, will be excellently answered by an AI Agent.

This is real. This is happening now. When an AI Agent can answer the vast majority of customer questions, this fundamentally changes the nature and makeup of a human support team. The work changes, the economics change, the metrics change. Nothing is left untouched. With AI-first Customer Service, support teams work on higher value, more complex customer questions, and they invest most of their time analyzing, managing, and improving the AI system, so it can deliver even better results for customers.

What is AI-first Customer Service? 

At the highest level, AI-first Customer Service involves first seeing if AI can resolve the problem directly for the customer, and if not then seeing if AI can help the agent solve the problem.

AI-first Customer Service means customers will get:

  • Instant, excellent answers to almost all their questions, delivered by AI.
  • And when it isn’t instant, they get deep, attentive, personal, human service because agents have more time and have AI to help them.

AI-first Customer Service has three components that work together in a single system:

  1. AI Agent first: Customers experience AI first. Instant, accurate answers.
  2. AI Copilot first: Human agents experience AI first. Instant always-on help that comes to them.
  3. AI Analyst first: Leaders experience AI first. An AI Analyst proactively delivering insights to them.

AI-first Customer Service is both a product change and a mindset change.

Businesses that change their mindset and adopt AI-first Customer Service will deliver remarkable customer experiences. Those that don’t will get left behind. At Intercom, we’re pioneering this new way. Let’s dig into the details.

AI Agent first: Customers experience AI first – instant, accurate answers

When customers have questions and problems, they want answers and resolutions, as quickly as possible. Talking to the customer service team is a means to an end. The best customer experience is an instant, accurate answer, or immediate action taken to resolve the problem.

AI can already answer many questions and resolve many problems, and oftentimes better than a human can because AI is always available. AI answers instantly, works 24/7, and speaks every language.

As the technology improves, and forward-thinking support leaders learn how to best set up AI-first Customer Service, AI is simply going to get better and better at answering customer questions.

This transformation is already happening. We launched our Fin AI Agent in May 2023. On day one, the average resolution rate was 28% of customer questions. Less than a year later, the average resolution rate is 46%. Thanks to AI and feature improvements we’ve built, and our customers learning how to optimize their content and data, our Fin AI Agent went from resolving a quarter of queries to a half of queries in just one year:

Aside from the pace of improvement, what is striking is that customer satisfaction scores remain high. In fact, for some customers who have deployed Fin AI Agent, their human support satisfaction has increased, because their human team has more time to spend with customers on harder problems.

“For some, AI is already answering up to 80% of their customer questions, without a drop in customer satisfaction”

46% is the average resolution rate across our customer base. For some, AI is already answering up to 80% of their customer questions, without a drop in customer satisfaction. For the majority, it’s answering 30-50% with almost no upfront investment.

AI Copilot first: Human agents experience AI first – instant, always-on help that comes to them

When an AI Agent can answer the majority of customer questions, only the hardest ones get passed to the human support team. These questions require research, troubleshooting, and can take time.

AI is excellent at helping with these tasks. AI can ingest far more content and information than any human can, including all historical customer support conversations, and can offer suggestions and advice to help the human agent. This is the AI Copilot, a personal assistant for every single human team member. Copilot is ever-present, connected to all data and systems, and provides information and troubleshooting advice in real time.


Copilot removes all the heavy lifting of using slow, outdated systems, digging through many docs and old conversations, and makes team members faster and more efficient. Copilot also means instant ramp time for new agents, and other agents needing to know new things. Because the AI Copilot has access to all conversation history and all satisfaction scores of those conversations, it can provide agents with not just the correct answer, but the best answers to questions.

We launched our Fin AI Copilot into beta in April 2024, and early customers are already seeing up to 31% efficiency improvements.

Copilot also teaches itself to get better. It learns from what human agents do and don’t say, as all new conversations and resolutions get fed into the AI Engine. As Copilot learns what the right answers are to harder questions, it tells the AI Agent, so the AI Agent can answer next time. It’s an intelligent system that gets better the more it is used.

AI Analyst first: Leaders experience AI first, with insights proactively delivered to them

The transformation of customer service is not limited to customers and human agents. AI is also excellent at many of the things support leaders need to do, like querying data to understand insights and performance in real time or retrospectively compiling reports and charts, and training new human agents with real-time feedback and suggestions on best practices. And AI can do all this far faster than any human support manager or leader.

Just as human agents have a Copilot, support managers and leaders will have an AI Analyst to help them, to answer their questions, but most importantly to proactively bring them new insights which they can act on to improve service quality.

“AI will elevate the role of a support manager to be much more strategic”

For the first time ever, support leaders have access to knowing the quality of 100% of their customer conversations, because rather than rely on customers answering customer satisfaction surveys, AI can understand whether a question was successfully resolved, and understand the sentiment of that customer. The AI Analyst proactively provides a holistic view across all conversations and suggests clear areas of improvement.

AI will elevate the role of a support manager to be much more strategic. Rather than wrangle the day to day of support operations, they will design and optimize the overall AI-first customer experience.

At Intercom, we already have early AI Insights features live, with much more to come and an AI Analyst product in the market later in 2024.

Data and knowledge content will become strategic priorities

The AI-first Customer Service platform also has a data and knowledge layer that AI uses, learns from, and improves. This centralized hub includes internal and external data, and internal and external content sources.

The quality of the data and knowledge layer is foundational to the quality of support. Intercom customers who have invested in better data and better content have seen huge improvements in the resolution rate of their Fin AI Agent. In fact, on many occasions we have seen customers think that our Fin AI Agent is giving wrong answers only for it to become clear on investigation that the underlying content is out of date or wrong.

This is the power of an AI-first Customer Service solution on one single, seamless platform. It can easily surface suggested areas of improvement, and the system gets better and better. Because it’s built together, Intercom’s AI-first Customer Service platform improves much faster than one where different tools from different companies are stitched together.

AI-first Customer Service is better, faster, cheaper

AI is already transforming customer service. This is not speculation, the evidence exists right across our customer base.

Customer service will be one of the earliest areas to be completely transformed by AI, and the transformation will be incredibly positive.

The customer experience will be better. Customers will get instant, accurate answers almost always. Customer satisfaction will soar. First response times will be decimated.

When customers don’t get instant answers, they will get personal, attentive human service, where the human support team work on much more interesting problems, and have more time to give excellent support.

Customer service leaders will have much more strategic roles to play. They will work on designing a system that not only delivers excellent service, but that delivers new insights to the business about their customers.

And leaders will be able to reshape their team, moving people into new roles like improving content, AI system design, and doing much more with less. Support teams will be small and mighty.

This is an adapt or die moment for customer service

Customer service teams will adapt AI-first Customer Service soon or be left far behind. We already see early adopters reaping huge rewards with industry-leading customer service and soaring career progression. Some may think they are crazy at first, but they will win. Sitting on the fence is not an option. The pace of AI technology advancement is too extreme – it will quickly leave businesses that resist behind.

Intercom is pioneering AI-first Customer Service. Join us and we can build a better future together.

Fin AI Copilot horizontal CTA