Cisco Adds AI in MindMeld Buy
In Star Trek, when Spock wanted to get smarter about something, he could create a telepathic link between himself and another person. Called the Vulcan mind meld, the technique allowed for an exchange of data (thoughts) enabling the participants to act as one mind... which would hopefully lead to smarter and better decisions.
However, we're still a few hundred years from forming an alliance with our future friends. For now, we'll need to find a better way of doing things faster and smarter, and companies are looking to fill that gap through artificial intelligence (AI). Of all the unified communications vendors, Cisco has been among the most vocal regarding the use of AI for improving collaboration. Back in July 2016, Cisco previewed a speech interface, called Monica, and made it clear that AI was something it was pursuing.
Today Cisco fast-forwarded its internal efforts by announcing the intention to acquire a small (24 employees) startup, called MindMeld, for $125 million. For those not familiar with MindMeld, the company has an AI platform that enables businesses to build conversational interfaces into applications and devices. The company built its own proprietary machine learning (ML) algorithms that facilitate natural language transactions though voice and chat assistants.
In addition to talent, MindMeld will bring the following capabilities to Cisco:
- A conversational AI platform enabled by large-scale ML
- Advanced natural language parsing with capabilities in the areas of entity extraction, language semantics, and pattern recognition
- Extensive vocabulary libraries and knowledge graphs for quick answer generation on frequent questions
Cisco will use the technology to advance Spark, its team collaboration tool, in the areas of ambient listening. In his March Enterprise Connect keynote, Jens Meggers, Cisco SVP and GM for the Cloud Collaboration Technology Group, gave us a taste of what's to come when he prompted a Spark Board room-based team collaboration device to call someone through voice commands. Once the MindMeld integration is done, a group of doctors in a room outfitted with a Spark Board could ask the device to pull up a patient's health records and X-rays. Or a project team could ask Spark Board to find the most relevant information on oil and gas exploration, and the MindMeld ML algorithms would quickly do so. This lets companies tap into collective knowledge without requiring a Vulcan to perform a mind meld on every co-worker.
The low-hanging fruit for Cisco is to leverage MindMeld in integrating with its Spark platform in the following areas:
- Spark Bot -- A chat assistant that would interact with workers and execute tasks autonomously
- Spark Care Bot -- A virtual agent that would handle customer care instead of a human
- Spark Board -- Contextual voice or video interactions to enable more efficient meetings
- Spark Room Kit -- Enhance smaller meetings through audio and video
Since MindMeld is a cloud-based solution, the company is able to aggregate information from its customers. That's a good thing, because with AI, the more data and more experiences the system has, the more accurate it works. MindMeld gives businesses the ability to own and control the data and models while providing a base of data that gets them around a "cold start" period during which the AI needs to go through a learning phase before it can be come useful. Since MindMeld already has done that heavy-lifting, inference accuracy should be high.
There's currently a tremendous amount of fear and trepidation in the workplace regarding AI taking jobs from people, but I believe those fears are based on a complete misunderstanding of why we need AI today. AI will augment what we do, not replace us.
Earlier this week, I authored a post highlighting ServiceNow's new State of Work report, which found that workers today are drowning in data. In fact, 86% of the respondents indicated that, if things do not change, businesses will hit a breaking point by 2020 and sifting through the massive amounts of data collected will no longer be feasible. As has been documented on this site and many others, the Internet of Things is coming, and coming in a big way. This will simply add to the data tsunami headed into the workplace.
AI in the workplace can make us smarter and much more efficient. Instead of having to wake up in the morning and read through hundreds of emails and Spark messages, imagine if you could just ask Spark to summarize the data and tell you what requires immediate attention. In fact, it could perform many of the simple tasks for people, accepting or declining meeting requests based on calendar data, for example.
If you're of the belief that the amount of data in the workplace will continue to grow exponentially -- and who doesn't believe this? -- an AI capability accessed through speech and messages must drive the future of work. MindMeld accelerates Cisco's efforts in this area.
Looking into the future, it's easy to see how the technology could also be used to give IoT devices a voice. As endpoints become connected, they will become smarter and will interact with people directly and MindMeld can be used as the interface to create experiences that delight.
The acquisition is Cisco's fourth in 2017, and is expected to close in Cisco's foruth quarter, which is in the summertime. The MindMeld team will report into Meggers as the newly formed Cognitive Collaboration team.