cisco mindmeld open source FI

Cisco Turns MindMeld Acquisition Into Open Source Conversational AI Platform

Two years ago this week, Cisco acquired conversational AI startup MindMeld for $125 million. Four days ago, it transferred those MindMeld assets into an open source license complete with developer documentation and a GitHub repository. Cisco is positioning MindMeld as a Conversation AI Platform for enterprises. A Cisco developer blog post announcing the change commented:

“By making this open source software and companion documentation available to partners and developers, Cisco is committed to helping more enterprises find success building and deploying a wide range of conversational applications.”

A Conversational AI Playbook accompanied the announcement which offers examples on how to use MindMeld to build new applications. The playbook introduction summarizes the intent by saying:

“This playbook focuses on helping developers and data scientists build real production applications. The detailed instructions, practical advice, and real-world examples provided here should empower developers to improve the quality and variety of conversational experiences of the coming months and years.”

It is interesting to note that Cisco’s focus from the start has been MindMeld’s ability to tackle enterprise problems. However, the three example templates in the playbook are for consumer use cases such as video search, food ordering, and a home assistant.

Enterprise Conversational AI and Cisco’s Strategy

It is unclear about whether Cisco views its MindMeld acquisition as a success. Commentary around the time of the acquisition suggested the primary objective was to use MindMeld as a foundation for the Cisco Spark Assistant. That solution was announced six months after the acquisition, but other than a Webex assistant introduced in 2018, there hasn’t been much news about its AI assistant since then. Even without more products or revenue streams coming from the acquisition, the $125 million price tag may be viewed as a bargain to have an in-house assistant that you completely control with no licensing fees to pay to a third party. For a company the scale of Cisco, this may have been a good deal compared to the alternatives.

However, when a big company takes an acquired software solution open source, it is generally a recognition that they cannot generate significant revenue as a proprietary solution and they no longer want to shoulder the entire cost of sustaining the software. So, they transition to open source and hope independent developers or other businesses will take on some of the work sustaining the product. If MindMeld could have been positioned as an alternative to Cortana and Microsoft’s cognitive tools, then it had the potential to deliver a significant win and new product line for Cisco. But, conversational enterprise AI adoption is developing slowly and Cisco may not have the appetite to invest heavily in building out the market as well as selling a product into it.

MindMeld’s Future Will Still Depend on Cisco Investment

MindMeld as an open source platform has a lot of potential value for data scientists and developers so it may capture interest now that it is freely available. With that said, it is common for large technology companies to open source software and then generate little community support to maintain and extend it. Long-term success generally requires significant continued investment from the sponsoring company. So, MindMeld could become for enterprise applications what Mycroft would like to become for consumer applications, but only if there is a commitment from Cisco to building a robust, self-sustaining community. Mycroft has an independent business that is focused on building a community of developers. If you don’t hear anything from Cisco in the next six months about MindMeld, you are likely not to hear much about it again.

Microsoft Acquires Semantic Machines to Make Speech Engines More Conversational

Google Assistant Getting More Mobile App Integration and Control

Apple Acquires PullString as Voicebot Previously Reported – Here’s What It Means for the Voice Community