Today, Cisco announced its intent to acquire MindMeld, an artificial intelligence company based in San Francisco, for $125 million in cash. The deal is expected to close before the end of the year. The acquisition will be used to fuel advanced cognitive capabilities for Cisco Spark.
I’m on the road this week and have been unable to discuss live with anyone at Cisco. Here’s a few of my off-the-cuff thoughts on the transaction.
- MindMeld is less than six years old. TechCrunch shows the company raised $15.4 million in four rounds. MindMeld’s revenue is unknown, but I figure it must be low. However, ZDNet reports some high profile customers including Spotify.com.
- Without revenue the valuation isn’t clear. Though Cisco tends to evaluate acquisitions not on what they are, but what they can be. They look for technology that can scale. MindMeld has 10 patents. Cisco has a lot of potential use cases, customers, and sales staff to sell it.
- MindMeld was spun out of AI research at MIT.
- The 24 MindMeld employees will, like most Cisco acquisitions, be absorbed within Rowan Trollope’s organization. Specifically the team will report to VP Jason Goecke who knows something about the Cisco acquisition process – Jason got to Cisco when Rowan acquired Tropo.
- Last Fall, MindMeld released its Deep-Domain Conversational AI platform that enables its customers to create voice and chat assistants that demonstrate “knowledge and expertise around any custom content domain.”
- Cisco has been experimenting with AI for a couple of years now. Last June, it demonstrated Monica, a conversational personal assistant. Monica, or Cisco, has since gone mute. Cisco previewed a simple speech interface at the Enterprise Connect last March, but nothing conversational. There’s been some research with Alexa, Watson, and of course some automated text bots – but so far no conversational bot has been released.
- Amazon Alexa is the break-out star right now. Google has great technology, but no ecosystem. Google was an investor in MindMeld. Microsoft is struggling to get people to activate Cortana in Windows 10. If MindMeld works, this could be a key differentiator for Cisco Spark.
- What makes MindMeld compelling is technology that supports machine learning at scale, support for specialized vocabulary libraries, and natural language parsing.
- While several UC solutions have emerged that use Alexa – they still require an Amazon Echo device. Cisco has sophisticated microphone arrays built into its Spark Board and Room Kit systems for rooms. The new Room Kits also have some visual AI capabilities powered by Nvidia hardware.
- Cisco needs conversational AI to power things like Spark Bot, Spark Care Bot, and 1000 other other bots that will emerge.
- MindMeld may weaken the Cisco-IBM “partnership” which increasingly feels like smoke-and-mirrors to me. Originally, IBM needed a more contemporary option than its neglected Sametime, and turned to WebEx. That seemed like a simple win but WebEx begets Spark and Spark overlaps with IBM Workspace. That’s only a problem if IBM is serious about Workspace which is undetermined. Cisco will likely find some purpose for Watson, but hasn’t yet.
- I hope something more comes out of MindMeld than Cisco’s October acquisition of Heroik Labs, Inc. Heroik’s Worklife application was supposed to improve Spark with additional tools to “attract, retain, and engage users before during and after each meeting.” What became of that?
It is always sad to see a Star Trek inspired brand come to an end. I hope the technology will live long and prosper within Cisco.