Lenovo unlocks the value of generative AI in customer support

Using LLM-based AI technology to enhance speed, quality, and convenience for customers

Lenovo’s new AI-enriched, multi-lingual, omnichannel chat improves customer experience and enables faster and more accurate responses to support queries—while delivering a 15% productivity boost for contact-center staff.

Challenge: Staying Ahead through Innovation

As customers, we all get frustrated if we’re left waiting to speak to a vendor support agent – or if the agent doesn’t seem to understand our issue.

Positive experiences in the contact center drive loyalty, making excellence in customer support a critical factor in long-term business success. But for global businesses with large portfolios, it can be tough to deliver fast and effective 24/7 support in multiple languages—particularly given the cost of skilled resources.

Recognizing the potential for generative AI to transform the speed, quality and efficiency of customer support, Lenovo has created an AI-enriched omnichannel chat service for customers and agents. This case study explains how we delivered this innovative idea, what challenges we faced along the way, and how other businesses can learn from our knowledge to achieve similar benefits.


$1.3 trillion market for generative AI by 2032

Early adopters can increase competitive advantage

15% productivity gains in Lenovo contact centers


Same Goals, New Tools

“The basic goals in contact-center operations have changed little in the past ten years,” says Sourav Ganguly, AP Premier Support Director. “We still want to deliver the best possible customer satisfaction while maximizing operational efficiency.”

When the global pandemic accelerated the trend toward greater customer mobility, Lenovo recognized the need to offer more interactive and real-time support services. The latest Large Language Model (LLM) – based AI technology could enable the transformation by empowering human agents to work more efficiently and productively—while releasing them from low-value tasks to focus on more meaningful customer interactions.

However, transitioning from a legacy chat CRM system to an LLM-based AI CRM system implied a steep learning curve for both IT and business teams. Success depended on determining the most appropriate use cases, deploying targeted solutions, and rigorously evaluating their performance.

Brand image - Colleagues working together at a whiteboard.Brand image - Colleagues working together at a whiteboard.

Solution: Lenovo Powers Lenovo

Lenovo made a strategic decision to deploy Copilot for Dynamics 365 to enhance Premier Support contact-center productivity and customer experience with generative AI. The first project was an omnichannel chat capability in which customers can use one of nine languages to interact with the AI and Lenovo Premier Support agents.

The Lenovo solution automatically translates customer questions into English, suggests appropriate responses, and translates responses back into the original language. Premier Support agents review the suggested responses, and any edits they make are fed back into the system to improve the quality of content generation. Tracking edits enables Lenovo’s Premier Support team to benchmark the AI suggestion quality, and built-in sentiment analysis drives responses tailored to each individual’s experience.

SSG AI Pillars graphic: Security, People, Technology, ProcessesSSG AI Pillars graphic: Security, People, Technology, Processes

Four Pillars of AI Readiness

Pillar 1: Security

Avoided exposing sensitive customer data by paying close attention to cybersecurity and data policies. Copilot enforces role-based access control (RBAC) for agents.

Ensured safe use of Copilot through stringent processes: “Case summary” and “Ask a question” functions do not create new information but instead draw on qualified content and real cases.

Regular monitoring and dashboards track agent usage and adoption of Copilot-generated content. Agents review and send qualified content to customers, and all cases, including those with Copilot content, undergo regular quality-control checks.

Pillar 2: People

Gave business and digital teams time to understand the technology and its value, involved the business throughout planning and rollout in different geographies.

Used a closed-loop corrective-action feedback model to ensure rapid acceptance by agents. Ensured that agents see Copilot as a partner in achieving the desired outcomes and embrace changes in daily operations as opportunities to enhance their productivity and skills.

Pillar 3: Technology

Selected the right generative AI technology: Copilot provides assistance, guidance and support to users through automation and artificial intelligence. “For this very new technology, we focused on making sure the business got the most out of the existing Copilot capability, which gave us a great outcome, in a fast and easy way,” says Ziff Tang, Digital Transformation Support Leader.

Pillar 4: Processes

Accelerated post-call tasks by implementing automated summarization, leading to significant boosts in productivity. Additionally, enforced a governance process whereby agents review all generated content to ensure that it is appropriate for sending to customers.

Implemented quality management processes to monitor users across dimensions such as customer interaction, problem verification, policy implementation, solution effectiveness, technical skill levels, attitude, and responsibility.

Short Time to Value

Lenovo chose the top 5% most experienced Premier Support agents to design UAT scenarios and identify the most common technical issues and most effective troubleshooting steps. By focusing on business use cases rather than on fine-tuning the LLM, Lenovo was able to shorten the ramp-up phases and rapidly elevate the service experience of all agents. In just 45 days, Lenovo became the second biggest Copilot for Dynamics 365 user after Microsoft itself.

Previously, training for Lenovo Premier Support call-center agents was long and arduous. The new LLM-powered chatbot removes complexity, almost instantly providing high-quality suggestions. This AI assistance enables agents to look at the bigger picture and focus on delivering an optimized experience for customers.

“We’re really excited about disruptive potential of AI technology in contact-center management,” says Sourav Ganguly. “We are shortening overall turnaround time, improving diagnostic accuracy, and eliminating many non-value-added manual tasks.”

Brand image - Work Life. Woman in a conference room at work in a casual office environment.Brand image - Work Life. Woman in a conference room at work in a casual office environment.

Result: Raising Customer Satisfaction

As customers, we all want speed, efficiency, and accurate answers when we contact a vendor for support. Now that callers can use natural language to explain their issue, the experience is much smoother even before they reach a Premier Support agent. What’s more, Lenovo can gather  information more rapidly from each customer, so that agents are ready to deal with their issue right away. Full integration of information, combined with sentiment analysis, provides a holistic view of each customer to the agents and improves the consistency of responses.

“From day one, customer satisfaction levels were boosted up to 95%,” says Ziff Tang. “Every interaction is running faster, reducing wait times and boosting the Premier Support customer experience.”

Copilot draws on Lenovo’s rich history of 200 million service interactions per year to ensure accurate suggestions that drive customer satisfaction.

Cutting Average Handle Time

In addition to optimizing the customer journey, the AI solution has transformed work for Premier Support contact-center agents. Previously, agents would spend several minutes creating a detailed summary of each call and assigning a ticket. Generative AI now performs this post-call summarization within a few seconds, and the quality of the output compares favorably with that of the top-performing agents.

“The LLM-based chatbot provides a huge boost to productivity,” says Sourav Ganguly. “Even after the call, AI continues to improve the performance and quality of support received by our Premier Support customers.”

Average handle time for each contact dropped 20% after the introduction of the AI service, and it continues to fall.

Boosting Agent Productivity

During the initial rollout and ramp-up of the chatbot, the Premier Support contact centers recorded an average 15% improvement in agent productivity. In addition to striving for further efficiency gains, Lenovo is working on making the omnichannel chatbot more accurate by integrating the solution with its large internal support knowledge base. Using machine learning, Lenovo is identifying the most common customer issues and determining the most accurate solutions.

“Today, deploying generative AI is relatively simple,” says Ziff Tang. “However, if you want a best-in-class solution with the appropriate level of security and certainty that the generative AI will behave how you want it to, that is much harder. As pioneers in creating an enterprise-ready chatbot, we can support other organizations in designing, implementing, and optimizing AI-powered contact-center solutions.”


10% increase in customer satisfaction

20% decrease in average handle time

15% increase in agent productivity


To find out how you could benefit from Lenovo’s experience and AI-powered solutions to optimize your customer contact center, visit Lenovo.com or speak to your Lenovo contact.

Source link