All these devices, sensors, etc. that go in the IoT framework pump out a boatload of data. What are you seeing in IoT devices, and how are you absorbing all that data?
Sort by:
One of our customers is a company in Africa that actually uses our product to deliver blood using drones. There's no roads where they are, so drones are the only way to get blood out there. The general concept is: you're streaming it into the central repo and then you're leveraging your data science and your AI/ML to get it working and get some insights out of it. That’s why I joined DataBricks quite frankly. The need is pretty clear: having all the data is a big part of making good decisions around the signal to noise ratio… and pulling out what's different, what's important about this particular data point that's coming out of this, etc. I think that's critical. It'll be interesting to see if data itself becomes an indicator of whether the device is still considered secure. I think there's a lot of different ways we'll have to look at security in the IoT world, since it's hard to patch billions of devices that weren't designed to be patched in the first place.
There's a company called hunters.ai that’s partnering with Snowflake. They're taking the unlimited compute and storage power of Snowflake and creating a new, modern day SIM. To your point about the signal/noise ratio, there's no way you can look at that volume of data. You need smart intelligence to surface that information. It's always been true with some tools, but it's just that much worse now. The technology is there: DataBricks with their fast compute power and AI layer that can understand the data before you do. You won't be able to tune anything anymore. You need products that can self-tune. I do think it's a blossoming area.
I went to a private presentation a couple of weeks ago discussing how one of the differentiators is that they're pushing compute to the edge of the 5G network so that you can actually not have to home run to a data center to get computer resources to push it back. I think that may be part of what the drive between IoT in 5G in that relationship is; having the compute closer to the device, but in a broader and more mobile sense. I was also shocked at how much lower the power utilization is, ~10X less than what 4G is. Additionally, the device is small and it needs to consume less power because it is simpler to take advantage of the broader network capabilities. We do need to have better power consumption, so it'll be really interesting to see how this plays out practically.
Our carriers are progressing now, and exploiting the IoT scenarios (e.g., underwriters for insurance, medical malpractice, critical medicines that we need to deliver). There’s a diabetic monitoring tool that has an IoT device. It allows you to prick once, and then it’s constantly on. You can use your iPhone or your Android device to look at your blood sugar level. And there are a few other IoT based tablets and devices that are coming in. CVS Health is now expanding blockchain to that process. With COVID and the COVID medication that is going to come through, how do you make sure what you're getting is the correct one? So now they're exploring Blockchain and IoT to give some assurance to patients.
For us it's all about streams and then applying ML and AI to it. The underlying theme of AI is that it was figured out a while ago, but it didn't have enough data to work with. So now what we have to have is this compute and storage scale that allows us to leverage all the data. What our product provides to our customers is the ability to leverage data. How do you stream data into something in real time. How can you run AI/ML against it in real time? Our product is tailored towards doing exactly that.
The reality is it will take AI to understand what is happening. The volume of data is just too large. So the way forward is trusted data environments that you can look through with tools to understand what's happening in your environment. It’s just impossible, otherwise. That's why I think Databricks is going to be a really interesting play here in the future, as well as the products that people build on top of platforms like Databricks.