Agent.AI, a provider of mobile customer service technologies, today released its CRM software platform, which includes intelligent tools designed to streamline customer support interactions for live chat agents.
"Some of the older communication mediums—such as email and phone calls—are starting to veer heavily into messaging" platforms, says Fred Hsu, CEO of Agent.AI. "With the explosion of messaging, we thought it would make sense if there were better CRM productivity tools utilizing more automation, and really able to capture end users on their mobile and tablet devices."
Agent.AI aims to train chat applications, via machine learning and artificial intelligence (AI), to anticipate and suggest the best responses to a customer’s concerns and requests. The goal of this is not necessarily to replace human agents, says Hsu, but to reduce the time they spend tending to repeated processes, thereby freeing them up to address issues that require more considered attention.
In addition to dashboards that serve up customer data, analytics, session data, and purchase history to agents, the platform includes two core offerings: Co-Pilot, an intelligent chatbot designed to boost live agent productivity; and Auto-Pilot, an automated response system that can operate independent of inputs from human agents.
Co-Pilot processes past service interactions to learn, over time, the routes that reduce the number of steps it takes to reach resolutions. The technology can sift through a company's existing customer ticket resolutions, frequently asked questions, and customized events or interactions to better understand what the most logical response would be. Rather than spend time typing out reactions, users can review and send automated messages directly from an Admin Console.
The Auto-Pilot feature contains built-in "logic parameters" that make sure the responses are only sent out automatically when the system is confident in its decisions. According to Hsu, the system must be at least 90 percent sure the outcome will be positive before acting independently; otherwise, it will delegate the case to live agents. Hsu says that, on average, the system must analyze at least 5,000 messages before it can function to the best of its ability.
Hsu points out that the platform can help agents determine whether specific cases call for immediate attention. By reading data from past interactions, it can detect with natural language processing (NPL) capabilities whether a customer is getting irritated and communicate this via a separate bot. For instance, it will pick up on the fact that a customer has used all caps, profanity, or multiple exclamation points to get a point across. A bot can also assess data from previous transactions to determine whether the customer in question is a highly valued one to the company.
While Agent.AI can be used as a stand-alone product, it supports integrations with a variety of messaging venues, including SMS, chat applications (Facebook Messenger, Wechat, Line), email, and video chat applications like Skype. Users can also import third-party data from pre-existing technologies, such as CRM systems.
Agent.AI is available today. The first 5,000 messages are free to customers, and each following message is five cents.
“In the next six to 12 months, you’ll see us expanding a bit beyond just customer service,” Hsu says, to include e-commerce capabilities, for example.