chat bots

The recent shutdown of Facebook’s M, a human-assisted AI chatbot, has caused some people to write-off conversational AI. While it’s true that this was a major story and minor strike against AI, we firmly believe reports of AI’s demise are greatly exaggerated. At Glia, we’re still very bullish on AI and its role in facilitating meaningful conversations between companies and customers.

However, we still believe that in order for AI to be successful, the focus should be not on interfacing the chatbots directly with customers, but on interfacing the chatbots directly with the agents assisting them.

Some may ask, “Doesn’t that defeat the purpose of leveraging AI to replace human capital?”

To that question, we would say no because we feel strongly that the best implementation of AI is not to replace human capital at all, but to supplement it and make it better. The goal of leveraging AI in a conversational way should never be about replacing a person. It should be about delivering the best experience for customers.

When we released our agent-facing chatbot, OmniGuide, we did so knowing that we are still years away from having AI technology good enough to replicate a human-to-human conversational experience. While artificial intelligence will likely get there in time, the time isn’t now.

With a realistic appreciation of the technology and our mission to meet or exceed the in-person customer experience online, we believe our approach to AI is far superior – not only in delivering the best customer experience, but in delivering the most practical scenario for machine learning to occur.

OmniGuide was not designed to replace customer support agents. Instead, it was designed to empower them. For those unfamiliar with OmniGuide, our chatbot scans all web chat sessions with customers in real-time and provides agents with suggested answers to customer questions or actions that the agent can take to provide an optimal online experience for the customer. Agents can accept the bot’s suggestions with a single keystroke and pass those answers directly to a customer. If the chatbot’s suggestion isn’t quite right (which does happen), the agent can modify the response before sending or just downright discard it.

The beauty of OmniGuide is that every action the agents take is recorded and used to help train the bot to deliver a more accurate response in the future. We feel strongly that this is the best way to responsibly leverage AI at this time.

Focus on the customer experience and leverage AI responsibly. Think about empowering your agents, not replacing them.