🔊 Listen to this

The much anticipated Mary Meeker Slide Deck for 2017 from the Recode Code [1] conference is here. Last year Mary of Kleiner Perkins Caufield & Byers (KPCB) [2] established the foundation of what I have been calling the Voice First revolution.  Today she goes on to project the continuation of the massive growth It has seen over the last year.

Although the actual slide deck contains less direct Voice First slides, much of the commerce and company interaction insights on the other slides will inform the future growth of Voice first.

Here is a brief overview.  I will have a deeper analysis for subscribers of the Multiplex Magazine iOS app soon as a Multiplex Memorandum.  Not a subscriber? Subscribe! At the prices of a good cup of coffee per month you will get actionable insight about what is ahead and why it is important. Do it now, go here!

My early insights:

In slide 46 we see that a whopping 70% of Google in app searches are in natural language and not “search speak” this is an astounding shift that shows that we are rapidly shifting away from the legacy search systems and on to the new modalities of the Voice First dialogues. The slide shows a rapid expansion of 20% of mobile search queries are now via Voice First.  This is an astounding shift from mid single digits just last year.  The trend is due to double to 40% next year.

Slide 47 dives into the amazing early mover into Voice First, Amazon.  We can see since 2014 Amazon has expanded into growing Voice First into clear and identifiable segments:

1. Echo: Shopping + Audio Media
2. Echo Look: All Echo features + Shopping + Recommendations (Fashion and later general commerce)
3. Echo Show: All Echo features + Video + Voice Calling

We will continue to see Amazon splinter the use cases of Voice First systems as the entire market grows at a stunning pace.  The common elements will be Voice Commerce, but it will not be noticed overtly until it is rather late for others to respond in the market.  Of course the AI and machine learning will continue on a rapid growth and allow for deeper context whereby general AI would not nearly be needed as systems are tailored to you like a glove.

Slide 48 shows the dramatic rise in raw speech recognition.  Not to be confused with intent extraction, recognition is the ability to discern words for the intent extraction to operate on.  We are now past the line of average human cognition and although the line is flattening there is little more to go.  We will be at nearly 97% in months and thus we will have solved speech recognition to a greater extent.  The caveat is microphone technology. Amazon has championed far-field echo-cancelation, beam-forming 7-8 microphone radial arrays to solve this.  They have some of the best recognition in the world.  Google’s data actually would improve by a few points today if they established a reference platform for near-field and far-field microphone technology for manufactures.  Sadly the technology to make cell phone calls clear in may ways deter from clear automated speech recognition.  This is one reason Apple’s Siri at a distance in a room environment with any echo or noise is troublesome.  However AirPods are an amazing near-field system and the recognition is quite good and will be better after WWDC 2017.

We can see that almost nothing else is growing as rapidly as Voice First in Mary Meeker’s slides.  This healthy trend will continue to explode in 2017 and beyond.  Mary has had a keen eye and an outstanding track record of accuracy.  This is an amazing time for Google-class and Apple-class startups to launch, for I think there are dozens of these sized opportunities in Voice First.

Here is the record breaking 355 page in depth Slide Deck:

 

_____

[1] http://www.recode.com

[2] http://www.kpcb.com

 

~—~

~—~