DALLAS — AI-based assistants with which a use can communicate in natural language have become so commonplace in our consumer lives that it is believed by 2020, half of global searches will be done by voice. At its Partner Summit here, Cisco announced plans to add a Siri- or Cortana-style natural language “assistant” to its Spark collaboration system.

Dubbed Spark Assistant, the technology will at launch handle command and control — allowing users in a Spark collaboration session to make a call, start a meeting, record the session, and use other functionality in the company’s collaboration software.

Using voice, said Tim Tuttle, CTO of cognitive collaboration at Cisco “has been a huge change in behaviour that happens in the home, happens in the car — but my question was, why isn’t it happening in the workplace?”

While the “command and control” functionality demonstrated here at Partner Summit is certainly unique and interesting, it’s just the beginning of Spark Assistant’s life. Rowan Trollope, senior vice president and general manager of the applications group at Cisco, described five potential levels of AI in meetings and collaboration. Command and control is step one. The goal is something much more lofty.

“The real intent i to have a virtual team member who’s with you all the time,” Trollope said.

Trollope describes Spark Assistant eventually evolving into a full “thinking” team member in the collaboration room, doing the kinds of things AI are good at — always paying attention, never forgetting things. In fact, he described a future when Spark Assistant would jump in to keep a team on time and on subject, or to prompt input from a team member that hasn’t had much to say over the course of the meeting. Doing so, he said, would make meetings more productive, and would go a long way to help getting around one of the challenges faced by collaboration tools.

“You can have the best technology in the world, and still have a crappy meeting,” he said.

While the most common consumer AI personal assistants like Google Home and Amazon Alexa has massive user bases today, they are at their heart personal assistants — designed to answer the requests of a single person. Alexa, for example, doesn’t know that the voice telling it to play their music is that of its paired user’s teenage son, and thus will not respond with the youngster’s music. But that doesn’t work in an environment that is by its nature a team collaboration — Trollope said one of the biggest challenges they had to address was recognizing who’s speaking, and placing commands in the appropriate context. Fortunately, the company’s room-based collaboration systems in particular provide a very convenient set of tools to help Spark Assistant gather the clues it needs, in HD cameras and a wide array of microphones, as well as enough processing power to deal with the information gathered.

But ultimately, the company said users will not need a full room-based system to use Spark Assistant. The plan is to introduce Assistant on any and all systems that run Spark, from its desktop video endpoint, to PCs and mobile devices running the Spark client.

Assistant will be free to any Spark license that includes meeting functionality. So it’s not a resale opportunity per se, aside from the fact that as it develops, it may serve to drive more customers towards Cisco’s collaboration tools. But there is partner opportunity around Assistant, said John Graham, senior director of collaboration and software in Cisco’s global partner organization, in helping customers adapt their workflows for a voice-based environment.

Spark Assistant is slated to start showing up on SparkBoard and other Cisco collaboration devices early next year. It will be available in English at first, and then will spread from there. Tuttle said that by the end of 2018 it will be on “the languages Cisco customers care about the most.”