Want to double lead qualification? Learn how OUA did it with a gen AI agent 🤖

Register now


Empowering chatbot customer support with generative AI

image of the case study on chatbot customer support success

A publicly traded B2B SaaS company was facing efficiency and customer satisfaction issues within its customer service operations. They needed to reduce the handling time of common customer service inquiries so they expanded automation. Working with LivePerson’s conversational AI platform, Conversational Cloud®, the company was able to expand and update its customer support chatbot — boasting significant improvements to its overall support team and program.

The journey in numbers


deflection rate (up 30%)


first contact resolution (up 30%)


reduction in bot deployment time

50 NPS

for AI chatbot (up from -25)

70 NPS

for support program

Open the PDF case study for more, or keep reading!

Challenge: Inefficient customer service chatbots

This B2B SaaS company relied on automation in their customer support program, but their AI chatbot was cumbersome and making adjustments to the design and development took up to four weeks. This was frustrating because the bot was trained to recognize keywords but didn’t always understand the meaning and context of customer queries. These inefficiencies led to frustration from customer service agents who were spending too much time resolving issues that would be better served with automated support.

Solution: Implement generative AI in customer support

Chatbot customer support example using a B2B SaaS company customer interaction

Working with LivePerson’s Conversational Cloud, the company was able to improve its customer experience and the efficiency of its chatbot customer support with half the effort. Using the Conversational Flywheel as a framework, their improvements were focused on the Assist and Automate stages.


To better support their customer service teams, the company implemented LivePerson’s Conversation Assist — generative AI that is automatically trained on large language models and datasets and continues to learn and improve based on human agent interaction and feedback. With Conversation Assist, support agents were able to increase the speed and accuracy in which they respond to customer inquiries. 

The company also set up a Slack integration to help human agents and customer service chatbots better communicate with customers. This integration notified internal teams when bot service was needed — helping the team to resolve bot issues quickly.


First, the company turned on Conversation Autopilot, a generative AI tool that enables bots to automatically and dynamically generate contextually relevant responses by ingesting hundreds of support articles that generate detailed and natural responses without a large lift or setup. Next, they used Generative Intent Training to create training phrases that improve understanding of customer interactions and NLU. This helped the bot achieve greater accuracy and recognition for intent-driven questions. Lastly, they integrated the bot with multiple APIs so it could retrieve account-specific operation and other customer data metrics. These API integrations, combined with the LLM, freed up their agents to work on higher-priority activities.

Our customers are thankful for our automation resolving their most common inquiries within seconds. We receive sentiments such as ‘this was very helpful, and did feel like a natural conversation’ or ‘you are really helpful and can do anything I ask for, I really want to create a chatbot like you’. And what is even more rewarding, is that our support personnel can now focus on solving more complex problems, making their job more appealing as it provides challenge and growth.”

~VP of Customer Support

Results and outcomes

Today, this publicly traded B2B SaaS business has a fully functional customer service chatbot that understands customer intent and can be updated in half the time. This has helped their customer support teams improve their deflection rate and first contact resolution rate by 30% each. They’ve also improved their bot NPS from -25 to 50 in the first 18 months since adopting the Conversational Flywheel framework. All of these improvements have helped the support team bring their NPS up to 70. 

“Using LivePerson Generative AI and automation via APIs, we have achieved an automation deflection rate of 60% with a fascinating transactional NPS of +70!

~VP of Customer Support

See how LivePerson’s generative AI solutions can help your company