The world is alight with new features using ChatGPT. Our approach is to get the best of this incredible technology while steering clear of gimmicks and AI hallucinations. We’re focused on safely improving FAQ Bot with GPT to improve the customer and administrator experience. We’ve already announced FAQ Wizard - our GPT-powered content assistant– so content creators need significantly less time to create Q&A.
Now, we’re excited to announce our customer experience improvement using GPT.
Introducing GPT-enhanced conversations
This enhancement improves the ability of your chatbot to resolve queries, providing a better experience for your customers.
In the world of online customer service, giving customers the answers they need when they ask is king. However, customers often use one word questions or don’t repeat essential context when asking questions. This results in frustrated customers as they may not get the necessary answers. Our GPT4 enhancement to FAQ Bot is here to bridge that gap, interpreting the underlying intent behind customer questions.
So, how does this enhancement work?
If a customer asks a question and FAQ Bot doesn’t find a good match in your Q&A knowledge base, it calls on GPT. FAQ Bot provides GPT4 with the customer’s question, their prior recent questions, and information about your business and the bot’s role. With this information, GPT enhances the customer’s question - with the aim of finding a better answer match. If it finds one, it provides the relevant answer. If not, the usual processes apply, including web search fallback, contact form, or live chat, depending on your bot’s configuration.
An example from one of our users brings this all to life - the EVA Check-in bot. A customer asks the bot “how to integrate with teams.” With GPT4 context enhancement, the bot gets additional context that helps it understand the question and provide a relevant answer.
Is my content safe from GPT hallucinations?
You may be wondering, is this enhancement safe?
AI is used to enhance the question, with all answers coming from content you’ve approved in your knowledge base. So the only answers provided are those created and curated by you. We use the OpenAI API to power this feature, and any data submitted in the chat is NOT used to train OpenAI models or improve OpenAI’s service offering.
Chatbots already answer 80% of queries. While it’s too early to say what the incremental gain from this feature is, the early signs are very promising.
If you’re ready to give your customers a better customer service experience (and have less frustrated customers), start your trial here and get in touch to have this feature enabled (it’s not enabled on trials by default). Your customers will thank you.