As a customer support manager with many years of experience in the startup trenches –– from scaling global support teams at SmartRecruiters to launching conversational AI chatbots at Dapper Labs –– I’ve directly seen how technology transforms customer service.
Today, I lead CX efforts at Skybound Entertainment, where we’ve found success blending human ingenuity with smart automation. A key driver in this evolution has been the rise of AI customer service agents.
These virtual assistants are proving to be invaluable, enabling businesses to provide seamless support, achieve cost efficiencies, and ultimately delight their customers –– all while freeing up human agents for more nuanced interactions.→ Download Now: The State of Customer Service [Free Report]
Whether you’re a startup founder or a CX leader at a growing enterprise, this is your roadmap to leveraging AI customer service agents for support that scales.
Table of Contents
The State of Customer Service Report
Unlock essential strategies for exceeding customer expectations and driving business growth in a competitive market.
- Exclusive insights from worldwide CRM leaders
- Analysis of modern customer behaviors
- Closer look at the AI opportunity in CRM
- Strategies for staying agile in 2024 and beyond
Download Free
All fields are required.
What are AI customer service agents?
AI customer service agents are virtual assistants that use AI to interact with customers, answer their questions, and resolve issues in a conversational manner. They’re available 24/7 and can handle multiple queries simultaneously, making them an efficient and scalable solution for customer support. These agents utilize natural language processing and machine learning to understand and respond to customer queries, providing personalized and accurate responses in real time.
Why are AI customer service agents so popular right now?
AI customer service agents are a reaction to major changes in consumer and business operations — not just a passing trend.
From my experience at Dapper Labs, where my team and I launched three conversational AI chatbots using Ada — automating 70% (at the time) of incoming requests — I’ve seen how these agents can significantly reduce response times and improve team operations.
During that time, the use of the chatbots allowed for the team to work on clearing the large backlog while maintaining a steady rate of incoming requests. It was especially effective during new product launches and releases, which we had weekly.
These days, they’re not just chatbots spitting out canned responses –– they’re smart, adaptable tools that can handle everything from “where is my package?” to troubleshooting a tricky software bug and collecting the necessary information your team needs to deploy a fix.
Since Ai chatbots are built on LLMs and trained on your company’s data, they can deliver interactions that feel personal and human-like, deployable across a variety of your business communication channels.
Let’s go into why they’re blowing up this year, supported by data and my own experience.
Meeting Rising Customer Expectations for Convenience and Speed
Customers of today do not wait. According to a Salesforce report, 64% expect real-time responses, and according to Zendesk, 67% prefer self-service over speaking on the phone.
AI agents satisfy this need head-on, providing instantaneous responses wherever and whenever customers need assistance –– something human agents can’t really do without increased global headcount.
(PS: For a deeper understanding of AI-powered support tools, see HubSpot’s AI Customer Service Software guide.)
Offering Personalized Experiences at Scale
I remind my team often that “customers want to be seen.” Giving them personalized experiences is one way to do just that. In fact, 65% of customers actually expect personalized service.
While your human reps may not be able to fill this need all the time, AI agents have the ability to shine here, pulling CRM data to craft accurate, tailored responses quickly.
At Skybound, we use customer data from our store purchases to optimize certain customer interactions with the use of AI, focused on assessing customer sentiment and providing responses that are sensitive to the customer’s immediate needs. It's like Netflix’s recommendation system, but for service: it’s personal, and it works.
Reducing Costs and Improving Efficiency
I’m very used to working in agile and lean startup environments. So naturally, I’ve always kept cost in mind. According to Gartner, by 2025 80% of service orgs will lean on generative AI to boost agent productivity. A McKinsey study also finds that 35% already use it to amplify efficiency.
At Skybound, I recognized an opportunity to optimize our ecommerce and player support operations. By improving standard operating procedures and introducing automation into the mix, I was able to scale our support capabilities without accruing significant additional costs.
This resulted in a more streamlined and efficient workflow, enabling the team to handle increased volume while maintaining a high level of service quality.
Alleviating Burnout and Improving Employee Satisfaction
Too many skilled customer service reps get burned out from doing the same things over and over again. I’ve seen this happen and have worked hard to change it. A lot of the time this results in lower morale and leaves little time for important, complex work.
AI is making a big difference in this area. ServiceNow’s AI-powered automation has reduced the time needed to handle complicated cases by 52%, which means that human agents can focus on customer interactions with more important outcomes. I’ve seen how this type of change makes people happier at work, less frustrated, and eventually better for customers.
However, AI isn’t a magic bullet –– it needs to be used carefully. According to a study from the Institute of the Future for Work, “29-34% of workers felt more stressed when AI was used to spy on them instead of helping them.”
I’ve always pushed for AI to be a tool, not a watchdog. When used properly, AI improves both employee happiness and overall service experience by freeing up human reps from repetitive tasks and giving them more power to take on higher-value interactions.
It’s simple: when workers are happy, they provide better service.
Ensuring Consistent and Scalable Service
Many businesses find it hard to grow their customer service departments without lowering the level of quality in their work. Using AI customer service agents has become an important way to solve this problem and the market is growing very quickly.
AI in customer service, specifically, is expected to grow at a rate of 25.8% per year, rising from $12.06 billion in 2024 to $47.82 billion in 2030.
Smart brands are using AI to handle the mundane stuff –– those repetitive questions and basic processes that eat up valuable time. This frees up their human service agents to tackle the harder challenges that actually need a person’s attention.
And here’s the thing: major companies worldwide aren’t just experimenting anymore. They’ve cracked the code on AI assistants that work “around the sun” without making customers frustrated.
I think it’s proof that you can actually grow bigger AND better when you’re smart about blending AI and human support.
The State of Customer Service Report
Unlock essential strategies for exceeding customer expectations and driving business growth in a competitive market.
- Exclusive insights from worldwide CRM leaders
- Analysis of modern customer behaviors
- Closer look at the AI opportunity in CRM
- Strategies for staying agile in 2024 and beyond
Download Free
All fields are required.
Benefits of AI Customer Service Agents
The benefits of AI customer service agents are becoming increasingly evident, offering a transformative impact on how businesses interact with their customers.
These aren’t minor tweaks or gradual improvements either — they’re game-changing advancements that are reshaping how we operate and do business. In fact, HubSpot’s own VP and CMO weighed in with their thoughts on AI agents in Are AI Agents Worth It?
Now, let me share some more insights from my experience, supported by more data, that highlight the direction we’re heading.
Instant, 24/7 Support – Around the Sun
These days, in this digital world that is always on, you can’t just wait around for help. AI agents are properly placed to give customers the instant gratification they want and need.
A huge 65% of companies plan to put more money into AI for the customer experience this year. Why? Because AI doesn’t sleep or need breaks, and it can handle a lot of interactions at the same time.
Multilingual Capabilities
One of the best things about AI for customer service is that it can quickly talk to people from different languages. By providing correct answers in real time and in multiple languages, these tools are completely changing localized support.
Companies can now service customers around the world without having to hire large teams of multilingual agents. This makes international help easier to get (and much cheaper).
Improved Efficiency and Cost Savings
According to a 2024 Intercom study, 68% of support teams report their customers now expect lightning-fast responses thanks to AI. The bar’s higher than ever, and we’re all scrambling to keep up.
But here’s the thing –– it’s working.
About 65% of C-suite support leaders are hunting for AI tools to modernize their tech stack, and for good reason. The numbers don’t lie. Teams running with AI are crushing their KPIs and productivity metrics while decreasing cost and increasing revenue.
Here are some numbers showing you the impact AI has had on the various departments, reported by some organizations in a McKinsey study.
At Skybound, we’ve managed to scale our support operations without throwing more bodies at the challenge.
The result? We’re handling more tickets without watching our costs spiral out of control during peak seasons.
Declan Ivory, Intercom’s VP of customer support, says his advice is to “Move fast. Don’t lose out on the opportunity. It’s there for the taking (now).”
With 43% of teams seeing a direct link between meeting high customer expectations and keeping them around, AI isn’t just a nice-to-have — it’s survival in today’s business climate.
Customer Insights and Analytics
AI customer service agents are invaluable in making data-driven decisions. I’ve always focused customer support operations based on the discoveries I’ve found through general trends and patterns, as well as in-depth analysis.
Whether that be optimizing departmental efficiencies, team workflows, or directly impacting the product roadmap –– these insights become crucial for taking your business to the next level.
AI customer service agents give teams access to a wealth of data considering they can be connected across a variety of your interaction touchpoints. By proactively addressing these findings, whether through self-service resources or product updates, you can reduce friction in the customer journey.
Improved Customer Satisfaction
It should come as no surprise that using AI customer service agents can lead to happier customers. They want quick, accurate, personalized responses.
Optimizations that lead to improvements — such as higher first-contact resolution, lower customer effort scores, reduced average handling time, and increased personalization — have a direct positive impact on your customer’s experience.
The 2024 HubSpot State of Service report found that of surveyed customers:
- 78% said they expect more personalization in their interactions.
- 82% said they expect immediate problem resolution from customer service agents.
By using AI customer service agents, you can ensure you’re fulfilling these needs without putting too much stress on your team.
How to Implement AI Agents for Customer Service
- Define your objectives.
- Pick your tech (but choose wisely!).
- Prepare your data.
- Design your conversational flow.
- Train your agent like a pro.
- Connect your AI to your tech stack.
- Run a test focus group.
- Deploy and monitor.
Getting started with AI customer service agents can feel a bit daunting at first, but once you have a strategy in place, it becomes something of an operational exercise.
When I was building AI agents at Dapper Labs, my team and I would have daily optimization and weekly strategy sessions to continuously improve the AI from the build, launch, and post-release stages.
Drawing from my personal experience, here’s my step-by-step approach for getting started with AI customer service agents.
Step 1: Define your objectives.
Before diving in, you have to know exactly what you want to achieve. Maybe it’s reducing response times, getting a portion of routine questions handled automatically, or just making your customer generally happier.
When I tackled this with my team at Dapper Labs, we set our sights on a specific target: automating over 50% of incoming requests.
This wasn’t just a random number. It meant our support team could focus their energy on the tricky stuff that really needed their expertise. In our case, we were deflecting upwards of 70% of incoming requests during that time. Trust me — having a clear goal like this makes all the difference between just implementing another tool and actually transforming your support offering.
Step 2: Pick your tech (but choose wisely!).
Don’t just grab the first chatbot you see. Take a minute to think about what you actually need.
I like to fill out my own platform evaluation checklist whenever purchasing a new tool. It gives you some structure and a strategic rubric for evaluating your needs.
It’s really important to choose the right platform, one that addresses what you’re looking for and seamlessly integrates with your existing tech stack.
While there are many options, I encourage you to check out HubSpot’s Breeze Customer Agent. It’s robust, customer-focused, and best of all –– does not require technical expertise.
Step 3: Prepare your data.
Your AI customer service agent is only as good as the data you feed it. You’ll want to gather up some existing customer service conversations, support tickets, and internal documents.
Then, you’ll want to take the time to clean it up and organize the data so that you train your AI agent effectively. I always prefer to start these things right versus taking any shortcuts. Go through everything in great detail to ensure you’re working with reliable information.
Trust me, it pays off when your AI customer service agent is responding accurately, minimizing your ongoing back-end manual training.
Quick tip: I’ve found it’s better to have 100 solid, accurate documents than 1,000 messy ones that’ll just confuse your AI.
Step 4: Design your conversational flow.
Nobody wants their AI agent to feel like a question maze. You have to map out the back-and-forth interactions like you’re the customer interacting with your product. Think about every possible path your users might take, from basic questions to those “uhh, I need a human” moments.
If it helps, grab a pen and sketch it out. Flowcharts will help you spot any dead ends or awkward loops in the conversations before they frustrate your customers. Your goal is to make every interaction feel as natural as chatting with one of your human service agents.
When customers hit a wall and need further assistance, make sure they can smoothly transition to a real person with minimal effort.
Step 5: Train your agent like a pro.
Training is not a one-and-done deal. You’ll be doing initial training and ongoing post-launch training. In my own experience leading these projects, it’s helpful to set aside time for your team to meet, specifically for working on training and optimization. This includes revisiting logs where the AI customer service agent provided inaccurate information, did not understand a customer query, or led the user through some kind of loophole.
Reminder: You’ll want to update your AI customer service agent whenever you have new releases or product launches, even if they’re temporary. This is where having a small, dedicated team in charge of your AI comes in handy.
Step 6: Connect your AI to your tech stack.
Integration is extremely important to get the most from your AI customer service agent. You’ll want to connect it with your existing tools, especially your CRM and knowledge base. These will be integral in training your AI service agent and allowing it to have real-time access to your customer database.
By doing this, you empower your AI customer service agent to make those personalized experiences we talked about earlier.
Step 7: Run a test focus group.
Now that you’ve created your conversational flows, trained your AI agent, and connected it to your greater tech stack, it’s time to run some tests pre-launch.
Run through your conversational flows with actual humans. At Dapper, we did a couple rounds of internal testing with our customer support team, letting them “break” the chatbot so that we could identify those areas for immediate training.
Likewise, on a more public level, we’ve done something similar at Skybound using a test group with our customer loyalty community. Whichever approach you take, just make sure you test and iterate before launching.
Step 8: Deploy and monitor.
Okay, so your AI customer service agent is live. Now comes the tedious phase of monitoring and optimizing.
I’ve found you may have to do this a lot in the beginning. You’ll most likely have areas in the conversational flows you did not consider, and that’s okay.
This is the time to optimize through new training, creating new flows, and editing existing ones. Those daily and weekly training sessions I mentioned earlier will come in handy here. I like to think of this phase similar to working on a ticket backlog –– clearing the queue.
Make sure to:
- Track key metrics like resolution times, customer satisfaction, and the percentage of queries the AI handles independently (i.e., deflection rates).
- Pay close attention to customer feedback, both direct and sentiment analysis in your chat logs.
Remember, your focus should be on continuous improvement to maintain optimal performance and customer experience, especially if you have an update-heavy product or service.
The Future of AI Customer Service Agents
AI customer service agents are a major shift in how businesses operate and connect with customers. They offer fast, personalized, and scalable support, which the data shows is what customers expect today. My own experience shows these benefits are real.
Looking ahead, AI in customer support will only become more important. I think we’ll see even more personalized service, AI that predicts customer needs, and better teamwork between AI and human agents.
While researching and writing this article, I was reminded of –– and learned new data points around –– the rapidly growing customer expectations in the age of AI.
I recognized that customers now expect AI to enhance every aspect of their interactions. For those of us working directly with customers, it’s about anticipating how those expectations are evolving as AI becomes more prevalent in daily life. This is the new reality of customer service.
The State of Customer Service Report
Unlock essential strategies for exceeding customer expectations and driving business growth in a competitive market.
- Exclusive insights from worldwide CRM leaders
- Analysis of modern customer behaviors
- Closer look at the AI opportunity in CRM
- Strategies for staying agile in 2024 and beyond
Download Free
All fields are required.
![How AI Can Unlock Customer Insights [+Expert Tips]](https://knowledge.hubspot.com/hubfs/customer-insights-ai-1-20241101-2082956.webp)
![Predicting Customer Behavior and AI: My Journey to Understanding the Customer Mind [+ Pro Tips]](https://www.hubspot.com/hubfs/predicting-customer-behavior-1-20241119-9776241.webp)
