We know. Generative AI is mindblowing. But did you know you can solve real-life use cases with the technology instead of asking how long brown rice should cook? Shocking, we know. Let's prove it to you.

Customer care ft. ChatGPT

Long before OpenAI turned the world upside down with their generative AI tools like DALL-E and ChatGPT, we started experimenting with the technology to improve our client's customer care process. Why exactly customer care, you ask?

With increased digital expectations and going through uncertain times, we noticed numerous companies ran behind on customer support questions. Naturally, this left customers frustrated or business opportunities unrealised by not responding to commercial inquiries. And on a human level, it creates a huge strain on the support team. So how does AI comes into the picture?

Glad you asked. Leveraging the recent state-of-the-art technology makes a customer care agent's workflow more efficient, without losing the personal touch. Which in turn, of course, increases customer satisfaction as well. 

Let’s show you how. In this demo, we don’t focus on customer interaction, e.g. through a chatbot. But rather on how questions are dealt with by the support team.

Demo time

Phase 1

A question comes in. Our support model has been trained on existing documentation, FAQs and historic support tickets. This makes it possible to cross-reference the incoming question for historic similarities.

Phase 2

Our system proposes some answers to the support agent. They can then choose to provide this reply to the customer...

Phase 3

... or improve and add the reply. This feedback in turn improves the model for future support cases.

Phase 4

Using Generative AI (GPT-3), the model can create new and better replies tailored to the specific question and build on earlier replies.

Phase 5

We use the support agent's implicit feedback to continuously build a personalised knowledge base and fine-tune a custom Q&A matching engine.

Putting it into practice

We have built an up-and-running version of the scenario above. At In The Pocket, we integrated this workflow into Slack to augment our internal office & HR support. Slack is ideal because it integrates easily, and the chat interface allows for easy interaction with the model.

If you’re curious about how we combine state-of-the-art technology with a laser focus on providing value to users, give us a shout.