Welcome to Lex: AWS’s AI-as-a-platform-service
Tech firms threw AI code and services at devs in a further attempt to secede and ultimately tie them into their respective ways of talking to machines.
Facebook open-sourced its Caffe2 deep-learning framework – open-sourcing of code being a proven way in tech of increasing uptake by removing legal and licensing hurdles for the all-important, canary-in-the-coalmine developer demographic that helps drive early adoption and toe-hold acceptance of new tech.
Microsoft updated its Cognitive Services with the addition of three APIs – Face, Computer Vision and Content Moderator through its Azure portal. The APIs let computers understand – supposedly – a picture, to compare faces and group them, and to quarantine images – something Facebook might want to tackle given its problems.
Cloud giant AWS sought to leverage the popularity of its parent’s digital assistant, Alexa.
Amazon’s cloud business released Lex, an artificial intelligence service that lets you build software capable of taking human instruction by voice and text.
Lex employs the artificial speech recognition and natural language understanding capabilities of Amazon’s Alexa and – in theory – lets devs build applications running on top of AWS’s machine-learning framework.
You can, again in theory, build voice and text interaction and understanding without also building the deep, machine-learning framework that must accompany it.
To build their conversational applications, devs would give Lex sample phrases while Amazon’s service builds for you the machine-learning models that parse phrases, understand the intent, manage the conversation and produce output.
Like it or not, hype or future fact, AI and ML are big for tech vendors. We are living in times akin to the early years of the personal computer during the 1980s and 1990s, a Gold Rush era with many initial players.
During those years, the competition was whittled down through successful technology, partnership and marketing, through bad leadership, bad ideas and bad execution.
The firm that dominated personal computing offered devs a compelling platform proposition and the right tools.
That firm was Microsoft, the platform was Windows and the tools were, well, dev’s tools.
Here we are again.
Then the platform – the computer operating system – was king and tech firms played to lock devs into theirs through the apps that were built using their tools.
Today, everybody is playing it a lot looser: the platform is no longer a PC operating system, it’s a collection of servers and services underpinning each firm’s cloud.
The wisdom has it that the apps must be free-ish to roam. It’s as self-serving now as it was 30 years ago: to build the biggest and best supported underlying platform; only the apps are following the people.
The apps will suck in the data and the AI and ML interactions made on the endpoints by people. Those endpoints are other tech firms’ devices or clouds.
Lex is a platform built on AWS, and you can use it to publish apps for Facebook’s Messenger, Slack, Twilio, web applications and IoT devices. Lex handles the authentication and – according to AWS – scale, thanks to its cloud’s elasticity.
You just better hope AWS is having a good day.
AWS this week listed US financial services giants Capital One and Liberty Mutual and VoIP service Vonage as Lex users.
Microsoft Cognitive Services, while built for Redmond’s underlying Bot Framework, works on iOS and Android as well as Windows. It would have to given Microsoft’s poor showing in mobile and devices and the success of the other two.
And Facebook’s worked with chip makers Nvidia, Qualcomm and Intel in addition to Microsoft and Amazon to tune Caffe2 for a variety of mobile devices – iOS, Android, Raspberry Pi – and Azure and AWS clouds.
If you’re a coder, you should feel flattered by the attention. You should get used to it – expect more – at least while the big names scramble for control.®