“Hi, I’m Tammy! Can I help you?”

Those little chat windows that pop up in the bottom right corner of web browsers, particularly on e-commerce sites, are quickly becoming the Microsoft Clippy of the ‘10s. “Tammy” and her ilk promise to help answer product questions, find the right size garment, track shipments or guide customers through checkout. But most of the time, these bots can’t do much more than answer a few simple queries with short, canned responses.

Chatbots need an upgrade, and the good news is they’re getting one. Technologies like artificial intelligence (AI), augmented reality (AR), virtual reality (VR) and more robust network connectivity are bringing today’s robotic, web-based chatbots into the future—a future that promises to advance many-a-company’s bottom line with the help of human connection.

“Chatbots of the future will be able to facilitate deeper human computer interaction by providing a level of dialogue and awareness that will make surfing the web, seeking help from customer service or ordering goods as natural as a conversation you might have with a family member over dinner,” says Eric Powlesson, senior strategist at Seattle-based digital agency POP.

The Chatbot Will See You Now

One of the most visible examples of the movement towards Chatbot 2.0 comes from Babylon Health. The U.K. startup created an ambitious chatbot for the medical field, now in a pilot program with the U.K.’s National Health Service. The Babylon chatbot is designed to reduce the load on Britain’s 111 medical helpline, guiding users through a series of questions that can diagnose non-emergency medical conditions, suggest medications and refer users to a live physician if needed. The company says Babylon can correctly diagnose about 80 percent of the maladies that primary care physicians encounter.

Unlike the Tammys of the world (hardwired to answer certain questions a certain way), Babylon’s chatbot is powered purely by AI, incorporating 300 million pieces of medical information, which the company says is the largest single collection of medical data in the world. The algorithms programmed into the chatbot are designed to improve its recommendations over time, as it learns from the results of its diagnoses.

“With advancements in machine learning, bots will get smarter [and gain] an ability to remember and predict unique needs and intentions based on previous interactions with the technology,” Powlesson says of the implications AI will have on the advancement of chatbot technology.

Putting a Face to a Bot

Across industries, chatbots promise to improve the seamless (and personalized) digital experience people have come to expect today. Functionality and accuracy, a la Babylon, is an important first step. The next will bring chatbots out of data silos, enabling them to process a much more diverse, heavier load of information. In the future, chatbots will likely take the form of personal assistants that users interact with regularly.

“Contextual information is what will keep chatbots from being the next Clippy,” says Mahi de Silva, CEO of AI-driven chatbot platform Botworx.ai.

So how can chatbots get closer to personal assistant and learn this contextual information? More data and emerging technologies. If existing versions of these virtual assistants are capable of pulling from a multitude of data points, one way to make them better is to add to those data points using other technologies.

“Including AR or VR devices that allow for the collection of even more external data may prove very useful in helping chatbots collect data regarding human actions and interactions in real life,” says de Silva. “As those technologies continue to be enhanced, it’s possible that bots can begin to utilize things such as facial recognition and speech pattern analysis to understand when someone they are engaging with is experiencing certain emotions and can adjust the conversation accordingly.”

De Silva also envisions a future where chatbots can understand human gestures, allowing motion to be used as a sort of interface. Of course, these technologies will need a robust network behind them to ensure they’re actually useful as real-time assistants.

“A powerful network is an absolute necessity,” says Powlesson. “Because of the volume of chatbots in existence, and their reliance on a firehose of digital inputs like non-verbal cues and facial recognition, there will need to be a strong and consistent backbone for that data to flow across in order to ensure and improve the efficacy of what these chatbots can do.”