AI-Powered Customer Service: 10 Real-World Examples

Artificial intelligence (AI) has become a crucial tool for meeting consumer expectations, particularly for customer service tasks. In early 2022, more than half of the respondents in a Gartner survey said they were already using conversational AI—software that can hold up its end of a text conversation or phone call—for customer interactions. And that was before chatbots started to talk a lot more like people.

Large language models (LLM) like OpenAI’s ChatGPT and Google’s Bard write as naturally as we do. Add text-to-speech (TTS) technology and they’ll speak, too. Maybe they’ll take over most of your customer interactions soon. (Gartner predicts that a quarter of companies will make chatbots their first-line customer service channel by 2027.)

For now, however, these chatbots are still prone to hallucination. That’s the confident spewing of nonsense, which makes LLM chatbots risky brand representatives.

“We have lots of work to do on robustness and truthfulness,” OpenAI CEO Sam Altman tweeted shortly after the company launched ChatGPT.

That said, the LLM arms race is moving so quickly that the hallucination problem might be solved by the time you read these words. Indeed, AI chatbots may very well change the way you operate your contact center—and soon.

Consider: OpenAI released the first public version of ChatGPT in November 2022. Just three months later, Juniper Research analysts predicted that AI chatbots would take over as much as 70% of all customer service interactions by the end of 2023. If these analysts are correct, we’ll plunge into an utterly transformed customer service ecosystem any second now.

But chatbots are just one way AI can help enterprises improve customer service outcomes. This technology can also do everything from deeper market segmentation to automating back-office tasks. Real-world use cases often best illustrate a developing technology’s potential, so here are 10 examples of companies using AI for customer service—and, yes, some of them do use ChatGPT.

AI Customer Service: 10 Real-World Examples

It’s tough to separate the practical realities of AI from the hype, and AI generates a lot of hype. (For example, an introduction to the latest version of ChatGPT inspired New York Times columnist Thomas L. Friedman to quote Arthur C. Clarke: “Any sufficiently advanced technology is indistinguishable from magic.”)

Much of that hype is probably earned. But it doesn’t tell you how AI can help your customer service department perform right now, today. For that, look to the companies who are already using the technology to get results. These 10 stories paint a picture of the current state of AI customer service.

1. Hyperpersonalized Email Engagement: The Muse

The Muse is a major career site for younger job-seekers. While email outreach is a crucial marketing channel for the site, it also provides a key customer service function, delivering curated lists of job postings and training opportunities to users.

The site partnered with AI marketing platform Blueshift to make sure their emails spoke directly to individual needs. This “segment-of-one” automation uses predictive analytics to predict user preferences based on more than their expressed interests: location, previous behavior, and professed skills play into the AI algorithm as well. Customers responded positively, effectively doubling visits to pages recommended in the AI-driven emails and scoring an email-open rate of nearly 60%.

2. AI Customer Service Bots: 1-800-Flowers

Artificial intelligence allows companies to automate typical customer service tasks, including core revenue drivers like product orders. Online floral dealer 1-800-Flowers worked with IBM’s Watson AI system to develop a “digital concierge”—an AI customer service bot that takes customer orders through their website and mobile app.

Titled “GWYN” (“Gifts When You Need”), this chatbot uses natural language understanding (NLU) and natural language generation (NLG) to take customer orders in a more intuitive way than a traditional online order form. “Instead of a structured process of filling out a form on a website, you’ll be able to just type into GWYN, ‘I’m looking for a birthday gift,’ and GWYN will ask, ‘Is it for a male or female? Age 30 or 35?’” Chris McCann, president of 1-800-Flowers, told Campaign.

3. Customer Sentiment Analysis: Flower Station

1-800-Flowers isn’t the only online flower dealer using AI to improve customer service. The UK-based Flower Station combines an AI chatbot called Tars with a sentiment analysis tool—MonkeyLearn—to “track and analyze customer feedback and reviews,” said Daved Cohen, CEO of Flower Station.

“This enables us to identify areas for improvement, address customer concerns

more efficiently, and make necessary changes to our processes” Cohen said.

The combination of AI tools has helped Flower Station improve customer satisfaction by 20%, Cohen said. Sentiment analysis alone—which uses natural language processing (NLP) to classify written and spoken responses—boosted positive customer feedback by 15%.

4. Branded Voice Assistants: Sensory Fitness

You’re probably already familiar with AI voice assistants; by 2025, analysts expect that more than 130 million U.S. consumers will use the technology, an increase of nearly 10 million users over just four years. Apple Siri, Amazon Alexa, and Google Assistant are well-known examples, but some brands are creating their own digital voice assistants to handle customer service tasks as well.

One example is workout education outfit Sensory Fitness. The brand developed a voice-powered AI assistant called Sasha to handle customer service phone calls. Like 1-800-Flowers’ chatbot, Sasha uses NLU and NLG to carry on dynamic conversations—but with the addition of text-to-speech (TTS) technology that speaks to callers out loud. The AI developer who created Sasha, FrontdeskAI, says the branded virtual assistant saves Sensory Fitness $30,000 per year.

5. Inbound Email Prioritization: Guardian Lemon Law

ChatGPT does more than chat directly with customers. California law firm Guardian Lemon Law uses the NLP tool to analyze massive email traffic—and prioritize the messages that need quick responses. That’s a crucial capability for a law firm that receives hundreds of emails per day, but lacks the staff to reply to all of them immediately.

“We need to separate the urgent from the non-urgent,” said Cameron Garrison, founder and CEO of Guardian Lemon Law. “Through language analysis, ChatGPT can recognize when an email needs immediate attention. That helps us work more efficiently, because it saves us both time and money.”

Since adopting ChatGPT for email classification, the firm noted a 12% increase in signed cases—a clear indication of improving customer satisfaction.

6. Conversational Interactive Voice Response (IVR) Systems: ING Bank Turkey

The conversational AI that powers Sensory Fitness’ Sasha can be woven into IVR systems to improve user experience during customer service calls. Turkey’s ING Bank found that its customer service team was overwhelmed by collections calls. (While the collections process isn’t the most positive customer journey, it’s certainly part of the customer experience.) The bank went searching for a way to automate this repetitive task, freeing agents for more valuable interactions.

The bank engaged conversational IVR provider Sestek, which uses AI technologies like NLU, NLG, and TTS to automate customer service interactions. With this conversational IVR system in place, ING can have complex, two-way conversations with customers, even during tricky collections calls.

The results? Agent workloads decreased by half—and customer payment promises soared by nearly 60%.

7. AI Support for Customer Service Agents: BOVEM

As we mentioned, LLM chatbots probably aren’t ready to take over customer service interactions entirely just yet. But ChatGPT is already helping agents work faster and smarter, according to Norman Teo, co-founder of grooming product brand BOVEM.

“We use a set of strict ChatGPT prompts to complement our existing customer response macros and templates,” Teo said. “With the use of ChatGPT prompts, our agents can focus on more complex queries and reduce ticket escalation to team leads.”

So what exactly is a “ChatGPT prompt?” It’s the question you ask the chatbot. The language you use to request AI text makes a big difference in the result. Consistent, well-written prompts ensure consistent and appropriate responses. In short, BOVEM is using ChatGPT to help agents decide what to say, and when to say it.

“Our ChatGPT prompts help our customer service agents maintain a consistent response quality and style across different channels,” Teo said. “This allows for a tightly clad response that also enables quicker response times, resulting in improved customer satisfaction.”

You can’t argue with the results. Formalized ChatGPT prompts helped customer service agents reduce response times by 30%. The company also tracked a 20% increase in customer satisfaction rates.

8. Predicting Customer Expectations: Gradient Insight

As an AI consulting company, it’s no surprise that Gradient Insight is always experimenting with new applications for the technology. The company’s customer service team uses AI chatbots, of course. Like Flower Station, they use sentiment analysis tools. But they also leverage AI’s ability to analyze huge datasets and predict likely outcomes—a technology called predictive analytics.

Predictive analytics can help companies guess what their customers want today—and what they’ll expect tomorrow. Gradient Insight uses this technology to “forecast customer needs and tailor our service offerings accordingly,” said Iu Ayala, the consultancy’s founder and CEO.

AI-powered predictive analytics also helps agents make personalized recommendations for each individual customer. That capability alone led to a 15% increase in revenue, Ayala said.

9. AI Speech to Text for Capturing Customer Interactions: Notta

Text-to-speech technology turns writing into lifelike spoken language. A complementary AI tool takes language in the other direction: speech to text, which automatically transcribes the spoken word.

Why does that matter for a customer service department? According to James Jason, co-founder and CEO of speech-to-text provider Notta AI, the technology offers surprising benefits.

“By automatically transcribing both video and audio calls in real time, we can efficiently and accurately address any questions or concerns customers may have,” Jason said. “Transcriptions have been a valuable tool for quality control. By reviewing the transcripts, I can identify areas where I can improve and ensure that I’m providing consistent and accurate information to our customers.”

Notta also uses these transcripts to train new customer service agents.

“We use the transcripts to create training materials and provide examples of successful and unsuccessful interactions, so that our team can learn from real-life scenarios,” Jason said.

These AI-generated transcripts also help Notta’s customer service team get better and better, said the CEO.

“We’ve been using transcripts for analytics to track customer behavior and trends,” he said. “By monitoring customer inquiries and complaints over time, we can proactively address common issues and make improvements to better meet the needs of our customers.”

10. Brand-Owned Voice Commerce Applications: bol.com

In 2021, Dutch e-commerce superstore bol.com launched into voice commerce and AI-powered customer service. Their Google Assistant integration is ahead of the curve: Not only does it answer questions, share daily deals, and update buyers on their orders, but it does so with a unique, branded TTS voice. This voice was itself constructed using deep neural networks (DNN), an advanced form of machine learning that is a subfield of AI.

“We launched [the voice interface], and we did not love the robot voice we had with Google Assistant,” Vera Rensink, strategy and business developer at bol.com, said in a webinar exploring the use of custom branded TTS voices. “That’s why we got in touch with ReadSpeaker.”

Artificial intelligence contributes to ReadSpeaker’s construction of custom TTS voices, too; their VoiceLab uses deep neural networks (DNN), an advanced form of machine learning, to construct all-original neural TTS voices for brands and creators. In the case of bol.com, that turned out to be a lifelike male voice that sounds trustworthy, kind, and helpful; core traits of bol.com’s brand.

Learn more about this instance of AI-powered customer service here, or visit ReadSpeaker for details about custom branded voices in AI customer service.