IBM PowerAI — our Elevator Pitch



An elevator pitch explains your product and the client value your product brings in a way such that any listener can understand it in a short period of time. The historical reason it’s called an elevator pitch is because you should be able to present it to others during a brief elevator ride (under 30 seconds, unless you live in a city like Manhattan where it could be 60 seconds -:) ); a more modern explanation is that people’s attention span is very limited today and they expect you to get right to the point asap.

More broadly, you should be able to ‘pitch’ the overview and value of your product to clients very quickly. If you can’t do that something is probably not right.

Happily, the elevator pitch for the product I work on, IBM PowerAI, is really easy to give. We’re all about increasing both the speed and accuracy of AI, specifically deep learning and machine learning AI models. I can give that pitch in 15 seconds, so i exceed the ’30 second elevator pitch goal’ by a large margin. Since time is our most precious asset (you can’t purchase any more of it), increasing the speed of something is often highly prized by almost all businesses. For example, many PowerAI use cases today are helping solve time sensitive issues, like diagnosing illnesses, new drug discovery, improving patient outcomes, real time fraud detection/prevention, etc.

PowerAI runs on IBM Power8 or Power9 accelerated servers. By optimizing the sophisticated interactions and connectivity between our IBM Systems CPUs, the GPUs, memory, I/O and FPGAs in concert with their usage by the PowerAI software stack, we can greatly increase the speed of training and deploying modern AI applications. Once we can increase the speed of AI applications (for example training Deep Learning models), we can then enable and environment where clients can further fine tune their AI applications and the amount of Big Data they access to improve accuracy.

Like i said before, PowerAI is all about increasing both the speed and accuracy of AI. I’d like to close my elevator pitch by taking a closer look at PowerAI by starting here https://developer.ibm.com/linuxonpower/deep-learning-powerai/

Source: Deep Learning on Medium