Google flexed its on-device computing might last week at its annual I/O developer conference. This should serve as a wake-up call to other technology companies whether they be consumer-focused or enterprise technology companies.
I’m still impressed by the fact that Google compressed Google Assistant’s footprint by 200x (from 100GB to 0.5GB). Similarly, other advanced Google product features such as Live Caption and Google Translate carry tiny footprints, making them ideal for the most inexpensive smartphones (thereby extending Google’s reach and facilitating the Company’s effort to “index the real world”).
It is significant that the machine learning capability and neural networks required to operate these product features reside on the phone. No need to run processes at the edge of the network or on remote servers. This means zero latency, better security and improved user experiences.
Recall when employees started to bring iPhones to work in 2007? This period kicked off the “consumerization” of enterprise technology. We are about to see Google raise the bar in similar fashion. If a customer-facing software application can theoretically operate “on-device”, it ought to be designed as such.
Reach me at email@example.com.