Going to Real Time

Going to Real Time

Does a true real-time corporation exist today? I would say no. I have never seen one. But they do exist in bits and pieces. Progressive Insurance has been experimenting with it. It set up almost real-time claims processing—you have an accident, a Progressive van comes to the scene, takes pictures, gives you an estimate, and gives you a check. The point is the whole thing stays in the system for a couple of hours instead of a couple of weeks or a couple of months. Another thing Progressive did was to say, "All right, we charge you for insurance based on where you live, among other things, but where you live and where you drive aren't really the same thing." So in a pilot program they put a GPS sensor in customers' cars. The idea was to charge more per minute spent in a high-crime neighborhood, for example. The sensor gives them the ability to charge you just for the risk you consume, when you consume it.

Companies like Dell, Cisco and Federal Express absolutely understand real time as an aspiration. I would guess, however, that none of these three would tell you they operate in real time. They would say they want to, and they are taking active steps to get there. They are saying that every time they encounter a lag in a process, they're going to do something about it. Let's look in the supply chain today, and our customer relationship management system tomorrow, and our production processes the day after tomorrow.

Why do I say they aren't there yet? I have a Dell computer. As far as I know, it gives me no way to give Dell feedback. They don't know how I like it, how it could do more for me. There is no feedback loop to Dell from the machine I'm using. Whereas every time Netscape crashes, it launches a little application that asks, "What were you doing when Netscape crashed, and will you send this information to Netscape?" Why shouldn't everything have a feedback loop like this if connectivity is universal? With this real-time information, Dell could update or patch my computer, but it could also learn from my experience and build that into its new computers, improving its product in real time and not waiting until my next purchase in three years to learn that I wasn't happy.

Here's another version of creating feedback loops. Imagine a driver is spinning his wheels on an icy road. He hits the Onstar satellite link button, and you see a technician looking at a computer screen with telemetry data from the car. The BMW 5 series has something like 148 microprocessors, and a lot of them are receiving sensor data. A moisture sensor tells the windshield wiper how fast to go, and ABS sensors show how the wheels are spinning. So with all this data, the technician hits a button and downloads via the satellite link a patch to the traction control software, and off the driver goes.

We can go a step further. You can get rid of the people and download the patch automatically. A really smart software agent looks for the right solution and breeds a new piece of software that suits the situation of the wheels spinning. Then another step: If we find something wrong with the Boeing 737 tail assembly, we don't fix one plane, we fix the whole fleet, right? Well, if we find a way to improve this one BMW, why not, since they're all connected, fix them all? So the company uses the satellite link to upgrade the software of every BMW. The company benefits because the product is upgrading itself continually. BMWs keep getting better from their experience in the real world, and the company doesn't have to go out and do a focus group to find out what happens on a slick road, then meet and plan and change its engineering.

This article was originally published on 11-02-2002
eWeek eWeek

Have the latest technology news and resources emailed to you everyday.