Andy & Kevin 2 Regression Testing & Automation

So you look at an enterprise, then they're sitting on hundreds of millions of dollars of spend just to bulletproof the apps they need to run their business on. And largely we're talking about a function inside an enterprise that, look, let's be honest, hasn't changed a heck of a lot in 15, 20 years. Largely using the same tools for the last 15 or 20 years, AI trundles along, suddenly reaches a escape velocity point. AI has been with us a very long time. In fact, advance has been a leader in AI for a decade now. But suddenly it's ready for prime time. So, so GENIE launched a week ago.

Why now?

Why does this matter?

You know, here's what we've been doing and again, it depends on who you talk to. If I'm talking to a developer developing only brand new applications, they say, well, why do I need this? We're going to do all this manual testing, we're going to release it. In fact, I talked to someone just a few days ago about that and they said, that's great, and you released it. And now it's in a pile of 5,000 other applications. Oh, well, that's a different group that does that. Go talk to the cio. So you talk to the CIO and they go, oh, you have no idea how big of a headache I have. And so tell me about the headache. Well, we have x amount of it spend billion dollars. 25 to 30% of that is involved in software QA that even the developers don't really know about because it's going on over here. Right. But it's a huge pile of money and a huge effort and very long delays, you know, weeks or months. And it's been now called in the modern software QA world, that started with Mercury and Rational. Let's go back in history a little bit. That's 1995. It's 1995 and at that point we had ways to write automation scripts and we had the advent of the recorder. And so if you look at end to end testing, we have two methods today. You have basically Selenium, which is writing by hand, I should say. You have manual testing, which is still the majority of the world. It's of that pile of applications. 70% of all end to end testing roughly is still manual. It's done manually. Nobody ever automated those regression tests. And that's what it's called, regression test. Right. Because I made a change in the application. It still has to work the way it was designed to. I have these written test cases that has to happen so you've got about 70% manual then most of the rest of the 30% is literally writing by hand. Scripts usually in Selenium could be in another language. Usually Selenium. And Selenium has been around. It's a 20 year old language. It is, it has not kept up with the advancements we've, we've, we've got in web applications, mobile applications, et cetera in my opinion. And it's, and it has a slow. Right. It has a certain basis of what you can do in general, someone who's really good can probably plow out 30 steps an hour. Interesting. But we've had recorders, in fact we had modern recorders for several years that can do easily four, five, six times that. Right. A recorder, a modern recorder with self healing, with script optimization etc and visual recognition, it can be five to ten times that even. Right. So we can, we can be 60 all the way to 6003-004005-00600 steps an hour. Really fast. Basically you're limited by the be the application. I can't record the next step without the application responding with the next step. And so that's 2, 3, 4 seconds and then I've got 5 or 10 seconds for the human to think about it and then they click on the next thing or execute the next thing or fill in the next thing and that writes a script step. Right. So, so practically I can't go faster than about one step every 10 seconds. And so you can figure out what that is. That's six steps a minute. And so you know, it's 300 to 400 steps an hour. That's as fast as a recorder can go. That's the limit. Then that'd be someone who's really good, who really knows the application. Okay. So if we're going to automate manual tests, we have to be better and faster than the recorder, otherwise people should use the recorder. So you have to look at the industry and say why do you have. We have a customer that has over a million test cases across their application that are manual. Say, why haven't you automated those? Well, the cost of automating them. Because it's a high cost person. Right. The cost of QA automation engineer and the cost of maintaining them outweighs the fact that we'll just run them as manual tests with lower cost labor. That's why. So what would happen if we obsoleted the recorder? Not only did so, the recorder obsoleted Selenium and still people use it. I don't understand why, but the recorder did not obsolete most manual testing, because manual testers are inexpensive offshore, they can be very inexpensive. Right. And so it still didn't obsolete the manual tester. To obsolete the manual tester, you have to be 10 to 100 times less expensive than the recorder.