5 Steps to AI Driven Testing

Artificial intelligence is top of mind for executives seeking game-changing efficiency boosts, especially in software development and IT operations. In those software eating the world domains, data is plentiful — but talent and time are dear. Spanning dev and ops, software testing is as promising a target as could possibly be. It’s a billion dollar bullseye, A.I. style — nothing artificial about it.

The good news is that testing is well suited for AI, as described in my February post titled An Ideal AI Use Case, which noted the two main reasons why regression testing – in particular – is made to order for AI. First, it is comprised of routine work tasks that require more persistence than creativity. Second, apps that need regression testing generate the big data from which an AI can learn. 

So regression testing is ideal for AI-based automation. How do you make it happen in your organization? That’s now straightforward. These five steps get you there.

  1. Fire up Appvance IQ: The industry’s first AI driven test automation system makes AI scripting easy. Then, its AI generated scripts feed a unified testing engine. In short, Appvance IQ is an out-of-the-box system for AI driven testing.
  2. Choose your target: Your first application-under-test (AUT) should be well suited for AI driven testing. For starters, choose a web app. Save your platform-specific apps (mobile or desktop) and your IoT systems for later. Your first AUT must also have production logs readily available. Now, web apps generate logs as a matter of course. However, accessing those logs typically requires that the AUT run behind your firewall, either locally or at an infrastructure provider like AWS. SaaS apps such as Salesforce or Workday don’t qualify, since their production logs are typically not available to you as a subscriber. Lastly, choose a popular app. That way the production logs will contain ample user activity against which to test.
  3. Grab a log: AI scripting is driven by the user activity recorded in production logs. So, you’ll want to grab a recent log with sufficient volume of user activity recorded in it. A week’s worth should be plenty. The log can be native Apache, or W3C, or from a log data manager like Splunk or SumoLogic. Then export a relevant and sufficient slice of the log in CSV format, including columns for IP address or session ID, method, host, URL and status. 
  4. Generate scripts: This part is very simple. Appvance IQ’s AI Scripting module uses the production log as a big-data source of learning, marries it up against an application blueprint that it builds and from which it learns, and then generates a comprehensive portfolio of regression testing scripts. How comprehensive? If your log breadcrumbed ten-thousand user paths, your script portfolio will contain 10K scripts. Bam, just like that.
  5. Execute your regression testing: Script portfolios from AI Scripting are fed into Appvance IQ’s test execution engine, first of all for functional testing. But why stop there? Appvance IQ is a Unified Test Automation system, meaning its AI generated scripts can also drive performance, load and even security testing, in addition to functional. So, your AI driven regression testing won’t stop at catching functional regressions. It will also make sure you don’t experience bug-driven regressions in performance or security. What’s that worth? Several more good night’s sleep per month, at the least. Disaster avoided, at most.

There you have it. Five simple steps to AI driven testing. What are you waiting for?

Perhaps a demo of what this looks like? Register for one here.