Inflectra's Plans for Leveraging Artificial Intelligence in our Platform

3-Nov-2023 by Adam Sandman Product News

As we discussed in this article - The Risks of Trusting Machines to Check Themselves - Artificial Intelligence is radically changing the software development and quality engineering landscape. At Inflectra we have been working on several different avenues of incorporating both artificial intelligence and machine learning capabilities into our product and platform. In this article you can learn more about our plans.

inflectras-plans-for-leveraging-artifical-intelligence-in-our-platforms-blog-image

Artificial Intelligence / Machine Learning in SpiraPlan

There are several different use cases that we are planning on addressing with AI in SpiraPlan, including both traditional large data machine learning and also the ability to leverage the power of generative AI to automate and streamline common tasks in the platform. We discuss each in turn:

Deep Machine Learning

One of the main benefits of using the SpiraPlan platform over other products and tools is the rich integrated dataset that we have in the system, For example, SpiraPlan includes information on product requirements, test results, quality metrics, code commits, CI/CD pipeline actions, product defects, customer issues, development metrics and most importantly, rich information on all the relationships that connect these items.

Currently SpiraPlan has a robust set of dashboards and reports that let you run reports and perform Business Intelligence (BI) activities to make decisions. What we envision is that using deep ML we can mine this data more proactively and use it to generate predictive metrics that allow you to address risks before they happen. For example:

  • Identify flaky tests automatically and flag as such

  • Analyze risks, code commits, test results to find predictive failures

  • Look for areas of most risk in each product / release

Some of these will be done natively inside SpiraPlan, others may use external tools and extensions.

For example, Inflectra is currently partnering with Angulon AI to create models based on sanitized SpiraPlan data repositories to be able to generate predictions that would:

 

  • De-bottleneck the process of application update release.
  • Speed up the regression testing cycle.
  • Improve feedback by highlighting most important problems fast.
  • Instantly select which test cases to run (test coverage).
  • Fail fast: Automatically prioritize tests so that problems can be identified early.

Generative AI

As described in this article we will be releasing in SpiraPlan v7.10 a new plugin for OpenAI / ChatGPT.

This new plugin harnesses the power of generative AI to automate the following common tasks in Spira:

  • Generate test cases and test scenarios for a specific requirement or user story.
  • Generate development and management tasks for a specific requirement or user story.
  • Generate BDD Gherkin-style scenarios for a specific requirement or user story.
  • Identify the possible business and technical risks associated with a particular requirement

Based on feedback from users, we plan on supporting additional use case such as:

  • Instructing ChatGPT to generate test cases, tasks, and risks for all the specific types defined in the product template. Currently, it just asks it to create "tasks," "risks," and "test cases" generically.
  • Adding similar functionality onto the test case details page where it will be able to generate the test steps for a specific test case automatically
  • Extending the functionality to the requirements and test case list pages so you can select multiple items on a grid and generate the test cases, tasks, etc.
  • Adding support for generating child requirements/backlog items directly from the requirements grid and mind-map views. 

Artificial Intelligence / Machine Learning in Rapise

Self Healing Tests

In Rapise, we have already released the first piece of machine learning functionality in Rapise 6.2 - the ability to have self-healing tests that use multiple recorded locators during the application learn phase.

Then during the execution of the test, Rapise can dynamically chose the best locators to identify the objects and thereby increase the resilience of automated tests, thereby reducing maintenance costs.

  • It was implemented as an initial use case in 2019 and has been useful to many customers

  • The current version has proved to be useful but limited due to the risk of false positives

  • We are working to investigating deeper models to see how we can improve its utility and reduce the risk of false positives

Rapise integrates with a wide range of different image recognition (OCR) technologies and we are using these to aid in the execution of tests where image to text interpretation is needed.

Assistance and Optimization

  • We are planning on extending the existing Rapise Toolbox to include an  AI-based smart helper that will increase the productivity of test creation

  • We will be adding AI based Code analysis tools to Rapise that will assist in finding duplicates, suggest refactoring and optimizations, similar to CoPilot by GitHub.

  • Using deep machine learning algorithms to analyze execution practices and behaviors to optimize playback (e.e.g use of Sleep vs. Wait, use of Global.DoSendKeys vs. DoSetText)

As we continue to refine the use cases and release the functionality, we will provide more information in the Inflectra product news channel. Stay Tuned!