The Voice First Abstraction Layers.

Reading Time: 3 minutes



Will Amazon Alexa stay dominate or does a startup have a chance?

Answer: Yes and Yes.

The Abstraction Layers

The AI infrastructure created by Alexa, Siri, Cortana, Viv, Watson and others are just that, infrastructure. Much like an Operating System (OS) there are rudimentary services in the OS (apps), but the true value comes with the development software. I call this the abstraction layer or layers. On PCs and on smartphones the app is an abstraction layer above the services of the OS.

Amazon calls elements on the abstraction layer Alexa Skills. These are early days and the modalities of Alexa Skills are not adequate for scaling. At over 5,000 skills, discovery, activation and other issues are paramount. For example, how do you know what Skill is available and what allows you to remember the arcane commands to activate it. Additionally how would a an app store work? There is no real app and if built correctly the app will continue to learn and grow. How much should it cost? What are the incentives to developers on the long term to maintain the app? There are far more issues to solve.

This is a huge issue with all Voice First systems and I have studied this for nearly three decades and have a number of solutions. These solutions must be in place for any long term ecosystem to develop and thrive. Current systems by all companies are flawed. This can be fixed but it is non obvious to even learned experts in this field.

The AI infrastructure is conflated with the apps that are built in the abstraction layers. Much like one may see an OS app like a note pad and conflate this with a full featured word processor. The OS may have services that can also compete with the abstraction layer.

We also tend to conflate the hardware with the modality of the Voice First OS. The hardware is just a conduit. The hardware could be a purposely built device, smartphone, computer, appliance, car, etc. The hardware will be everyplace and in no place in particular.

Finally we tend to conflate underlying AI infrastructure with something we call a personal assistant, and this is quite a misnomer. The personalized nature (that means knowing you deeply) of the current personal assistants is minuscule and is more related to an account profile rather then actual a true personal assistant in the proper context.



The Voice First Startup Opportunity

Thus just like the PC revolution and the smartphone revolution, the Voice First revolution startup opportunity is in the abstraction layer. Indeed the companies that build the hardware, develop the OS and in the case of Voice First, matin the AI will prosper immensely, the startup opportunity is to build in the abstraction layers.

At some point in time the AI available by Alexa, Siri, Cortana, Viv, Watson and others will to a greater degree, normalize. At that point the Voice First OS would become a standard utility, much like electricity in a sense. Universally all AI systems ultimately will perform:

  • Voice recognition
  • Intent extraction
  • Location Awareness
  • Time Awareness
  • Task Awareness
  • Semantic Data
  • Out Bound Cloud API Connections
  • Task And Domain Models
  • Conversational Interface
  • Text To Intent
  • Speech To Text
  • Text To Speech
  • Dialog Flow
  • Access To Personal Informations And Demographics
  • Knowledge Graph
  • Social Graph
  • Social Data
  • More

Along the way startups may or may not find the need to build their own fundamental Voice First AI and OS. There are no huge barriers as we are in the very early days of the Voice First revolution [1].

Once all Voice First OSs are uniformly similar in ability, the value in the abstraction layers grow. Startups have one of the largest opportunities the tech world has ever seen once this is fully understood. The fundamental opportunity is building application in the abstraction layer.

Additionally the abstraction layer can interact with a true personal assistants and personal agents. There are many reasons this is a more preferred modality rather than using the Voice First OS as personal assistants and personal agents.

The Next Google, Apple And Amazon

Google builds on the abstraction layer of the internet. The iPhone builds on the abstraction layer of the cellular providers. This analogy is close but not perfectly accurate. However we can see the value the abstraction layers create above a uniform utility.

Thus I conclude that in the abstraction layer will produce the next Google, Apple and ironically the next Amazon. It is possible for these companies themselves to become primary builders on the abstraction layer, however if you see what all these companies actually did to earn their revenue, they built a platform for other to build on:

  • Google: Platform knowledge graph, advertising
  • Apple: Products, apps
  • Amazon: Marketplace independent sellers, AWS web service users

There are not only opportunities for startups, there are Google, Apple and Amazon level opportunities.

[1] There is A Revolution Ahead and It Has A Voice




Of interest: