Snips to Enter Smart Speaker Arena and Fund the Effort with an ICO
The Paris and New York-based AI startup Snips yesterday announced plans to launch a smart speaker in 2019 and will fund the effort with an ICO (initial coin offering) of AIR Tokens. Snips was founded in 2013 and has developed a personal voice assistant that runs on device with no requirement for data to reach the cloud. To date, Snips has focused on providing an SDK for developers to add its voice assistant capabilities to mobile devices and Raspberry Pi devices.
The introduction of a smart speaker would deliver an entirely new distribution channel and potentially make Snips more accessible to non-technical users. Company CEO and co-founder Rand Hindi says the devices will be available in 2019 and will include language support for Spanish, Portuguese, Italian, Mandarin and Korean in addition to English, French, German and Japanese which are supported today. Snips also contemplates a device ecosystem connected by mesh networking that connects smaller devices with larger speakers promising higher audio quality.
A Focus on Privacy
Snips is attempting to differentiate itself by stressing privacy. Since all voice processing occurs locally on the device, there are no concerns about GDPR compliance nor any of your data being shared in the cloud.
Everything runs entirely on device. This makes Snips AIR the first true privacy-by-design voice assistant. Because what we say at home, should stay at home.
This may seem like a well-timed swipe at the recent Amazon Echo flap over the inadvertent voice message sent by Alexa from a couple in Oregon to one of their contacts. However, any solution that enables voice messaging between users could result in the inadvertent recording and sending of a short conversation. The more important distinction for Snips is that data processing all remains on device and is not shared with a cloud provider. Snips takes direct aim at Amazon and Google as aggregators that insert themselves in the process as a required intermediary that gathers consumer data about device use and interaction. The open source, on-device Snips solution offers a clear contrast to the proprietary, cloud-based solutions of the leading voice assistant providers.
However, that doesn’t mean Snips is alone in its current business and architectural model. Mycroft also provides an open source voice assistant with local processing, no cloud component and a community of developers. The company already has a smart speaker in the market and has a smart display scheduled to ship in 2018. With that said, Snips also stresses that it uses blockchain to distribute application content and conduct transactions which will ensure privacy beyond day-to-day use.
Funding by ICO
As if using voice, AI, blockchain and GDPR terms in the marketing pitch weren’t enough, Snips has also decided to add ICO to its cache of hot trends in technology. Crunchbase data show the company has previously raised $21.3 million and it is starting the process of promoting an ICO to raise much more. This is interesting since there is contemplated as a utility token application for users to acquire voice apps from independent developers and conduct purchases. More important for the company, ICOs have recently been used by several startups to raise tens and even hundreds of millions of dollars.
That is the scale of funding that independent AI startups with voice assistant and smart speaker ambitions appear to need to compete against Amazon, Google and Apple. SoundHound recently announced a $100 million venture capital round that followed a $75 million funding round just about one year earlier. Amazon has a reported 5,000 employees working on Alexa. Google has billions of dollars invested in its AI infrastructure. All of this reinforces the idea that building a voice assistant that works well for consumers is an expensive proposition. An ICO might be the best vehicle available to startups today to raise the funding required for success.
An Immensely Complex Undertaking
It would be easy to underestimate the complexity of the challenge Snips is undertaking. There is the complexity of building automated speech recognition (ASR) and natural language understanding (NLU) for multiple languages. Amazon is already seeing how hard it is to do this and match companies like Google that have invested in these capabilities for the past decade. In addition, you have the hardware innovation around far-field microphones for voice recognition and speaker sound quality. Then you have supply chain and distribution challenges for bringing the device to market.
It is likely that an open source voice assistant provider like Snips or Mycroft can carve out a role in the industry with the rising consumer interest in privacy and enterprise need for control. The success of Snips’ ICO and planned Kickstarter campaign will be good indicators of consumer interest in a smart speaker that promises privacy over functional breadth.