Questions about the AI and machine learning capabilities, including auto-healing. Select a question to expand its answer. algoQA has features called ReScraper and Locator Mapping to identify delta changes whenever an application undergoes changes. The Auto-Healing feature helps to fix the DOM changes. Auto-healing handles locator changes for small updates. Larger feature changes require scenario or step adjustments. algoQA accepts requirement documents, existing test case documents, and defect information to generate test cases that help identify coverage gaps. algoQA has impact-based testing capability to detect code coverage gaps. Additionally, algoQA-generated test cases can be mapped to requirements using tags to provide coverage insights. algoQA offers consulting and professional services for test automation implementation and optimization, in addition to its platform features. Unlike GitHub Copilot and Amazon CodeWhisperer, which are AI coding companions, algoQA focuses specifically on software testing and test automation. Its platform includes custom actions functionality and configuration options, allowing users to tailor their test scripts and extend the framework to meet their testing needs. Additionally, algoQA integrates with tools like Sikuli, Squish, and TestComplete for Coordinate-Based Testing, enabling testing in scenarios where traditional locators are not available or applicable. Prompting works in offline as well as online mode. The algoQA model needs to be created with proper features and nodes to use Prompting. algoQA leverages LLM models to find the correct locators to heal the scripts. algoQA stores the locator information along with the parent DOM information, and algoQA identifies the unique locators using this data. The auto-healing feature in algoQA, when a locator fails, automatically attempts to find a new locator for the failed element based on its properties such as text, class name, or tag name. This new locator is then used to continue the test execution. The process of finding a new locator is performed dynamically during test execution. To enable the auto-healing feature, follow these steps: 1. After generating the script, download it to your local folder. 2. Open the folder where you downloaded the script. 3. Locate the `ApplicationSettings.XML` file and open it. 4. Set the auto-healing setting to True. 5. Save the changes you have made to the application settings file. The following features of algoQA leverage AI: Scraper, Auto-Healing, Copilot, Codebot for creating custom methods, Jira Defects to Test case conversion (Test design), Prompting, Auto test case generation, deduplication, Feature file quality analyzer, and Smart Ordering. AWS Bedrock for LLM Deployment. Amazon Bedrock foundation models. Prompting leverages the LLM for test case generation, and the offline model is the input for Prompting. Proper input data is required for the LLM model, otherwise you have to edit the test cases based on requirements. Handling unstable or highly dynamic DOMs in applications can be challenging during test automation. One approach to address this issue is to use dynamic locators and WebDriverWait with expected conditions. - Dynamic Locators: Instead of using static XPaths or CSS selectors, use dynamic locators that can identify elements based on their properties or text content. This approach can help in handling elements with unstable or changing IDs or classes. - WebDriverWait with Expected Conditions: Use WebDriverWait with expected conditions to wait for elements to appear or disappear based on specific conditions. This approach can help in handling elements that are loaded dynamically or appear or disappear during test execution. algoScraper has the capability to get the optimal XPath for dynamic elements. Yes, algoQA's auto-healing feature allows the platform to adapt automation tests to match changes made to the application. algoQA supports various platforms for its auto-healing feature, including web services, green screen applications, and modern web technologies. For more details, refer to the "Frequently Asked Questions" document under the "Flexibility and Extensibility" section. Yes, algoQA's auto-healing feature can help adapt to application changes and update object IDs or locators as needed. This ensures that scripts remain functional even when the application's UI elements change. For more information, refer to "Auto Healing". Auto-healing in algoQA addresses minor UI locator changes automatically while maintaining traceability logs. Major application changes still require manual validation. If the button attribute changed to btn, then algoQA will be able to heal. Generated scenarios are designed to be semantically accurate and contextually relevant, not guaranteed to be perfect or immediately executable. Accuracy should be understood as how well the scenario captures the intended business flow and reduces manual effort, rather than whether every step is correct without review. Not in the traditional machine learning sense. The foundation model used in the algoQA platform does not learn or update itself based on iterations or feedback. Improvements across iterations come from better input, clearer intent, and more relevant context, not from the model becoming smarter. Iteration refers to refining alignment, similar to how a human tester produces better results when requirements are clarified. Each iteration helps the system by: - Adding missing context - Clarifying assumptions - Narrowing scope It does not change the underlying model. RAG allows the system to ground generation in relevant UI metadata and existing scenarios at the time of test creation. This improves relevance and consistency, but it does not introduce learning or memory into the model. GenAI models are probabilistic, not deterministic. This variability helps explore alternative flows and edge cases but means output should always be reviewed. This behavior is expected and by design. No, GenAI is intended to accelerate and assist test design, not replace human judgment. Final validation, domain reasoning, and intent confirmation remain the responsibility of the tester. Yes, LLMs are used to provide stable locators and assist with online recording of test scenarios using algoScraper. However, this can be bypassed with a minor performance degradation. AI is not mandatory to record automation scripts in algoQA. algoQA offers multiple approaches to test design. Test cases can be created using offline recording which is not AI dependent. AI-assisted test design using text inputs like natural language test intent prompts, defect descriptions, and manual test cases accelerates automation resulting in a significant productivity boost. AI features in the algoQA Platform: - algoQA-CoPilot: The integrated CoPilot uses AI to provide real-time in-product assistance, helping users resolve issues instantly. - Test Design: AI is extensively leveraged in the test design phase to generate test cases. - Code Generation: LLMs are used to assist users with extending default frameworks by suggesting code to create custom actions. - Optimization and Deduplication: AI assesses the quality of feature files, detects duplications, and helps merge similar tests for more meaningful automation. - Auto-Healing of Scripts: AI detects locator/label changes during execution and automatically fixes failed script runs, updating the object repository where possible. These AI applications enhance the efficiency, accuracy, and automation capabilities of the algoQA platform. AI cannot be disabled in SaaS deployment. For on-premises deployment, GenAI features are disabled by default unless specific conditions are met: 1. algoQA Bedrock Domain Whitelisting: Customer firewall or VPC must allow outbound HTTPS access to algoQA's Bedrock endpoints. 2. Provide Matching AWS Infra: A dedicated Bedrock-enabled AWS account/environment must be configured and accessible to algoQA. 3. Customer-Hosted Equivalent Models (Unsupported): While some customers may host equivalent models, algoQA does not support or maintain these due to variability issues. Key Considerations: - algoQA does not commit to deploying or managing LLMs on customer infrastructure. - If GenAI access is unavailable, relevant features degrade gracefully or remain disabled, without affecting core algoQA functionality. - Customer test data never leaves their infrastructure if using on-premises deployment with Bedrock access. AI is not required for offline recording. For online recording using algoScraper, LLMs are used to provide stable locators and assist with online recording of test scenarios using algoScraper. However, this can be bypassed with a minor performance degradation. If UI changes slightly, you do not need to re-record all scripts. algoQA employs auto-healing mechanisms to handle minor UI changes, such as locator changes or label modifications. When a script fails due to such changes, the agentic AI in algoQA updates the locator and re-executes the script. These updates are reflected in both the script and the object repository, ensuring minimal manual intervention. Yes, you can use algoQA purely as an automation tool first and add AI later. Yes, automation can be done completely without AI. While AI can accelerate automation and reduce manual maintenance through features like auto-healing, it is not mandatory for performing automation with algoQA. The tool can adapt to dynamic UI data changes through several features: - Auto-Healing: algoQA leverages LLM models to find the correct locators to heal the scripts. It stores locator information along with the parent DOM information, allowing it to identify unique locators and adapt to changes. - ReScraper and Locator Mapping: These features help identify delta changes whenever an application undergoes changes. The Auto-Healing feature then fixes the DOM changes. - Conditional Flows and Loops: Tests can handle conditional flows or loops by using the Record Scenario feature to record workflows for different dynamic values. This generates separate Feature Files for each condition or loop iteration. - Object ID Updates: You can update object IDs using algoScraper, re-scraping capability, auto-healing, or directly within the scripts. The Profiler feature allows editing the Object Repository file to apply changes to all relevant scripts. - XPath Mapping: For shadow DOM, iframe elements, or missing locators, you can update XPath expressions to target specific elements. The XPath Mapping feature allows replacing old locators with new ones. These features collectively ensure that algoQA can effectively adapt to dynamic UI data changes with minimal manual intervention. algoQA Copilot is a built-in chat assistant within the algoQA platform that provides real-time support to users. It helps by answering questions and offering guidance on using the platform. To access algoQA Copilot, follow these steps: 1. Log into the algoQA platform. 2. In the status bar of the algoQA platform, click on algoQA Copilot. A Chat Assistant window will appear. You can type your question directly into the text field and click the enter button to send your question, or select from the list of suggested questions displayed in the assistant. You can drag and move the pop-up window to different areas of the screen to view other parts of the platform while interacting with the algoQA Copilot. algoQA has an auto-healing feature leveraging agentic AI, prompting using Generative AI, Smart Report Analysis, and Smart Ordering features using AI/ML. The auto generation of code in algoQA is driven by user input and scenario specifications. Here is the exact flow of how it works: - User Input: Users specify the features they want to include in a test case. This input can be minimal, focusing on key functionalities or scenarios they wish to test. - Auto Generate Test Cases: The system uses advanced algorithms and machine learning methodologies to generate test cases based on the user-provided features. Instead of a "Record Scenario" approach, users directly specify the features to be covered. - Feature File Generation: The system automatically generates Feature Files that outline the test cases. These files are produced without requiring manual test design. - Review and Edit: Once the test cases are generated, users have the opportunity to review them. Users can make any necessary edits to fine-tune the testing process according to their specific requirements and domain knowledge. - Learning and Improvement: The system learns from the edits made by users during the review process. This learning is applied in subsequent test case generations, leading to improved accuracy and efficiency. - Leveraging Previous Test Cases: If users have previously recorded test cases using the "Recording Scenario" approach, the system leverages this information. It incorporates the knowledge gained from recorded test cases to enhance the quality of newly generated test cases. - Managing Workflows: Users can edit scenarios by modifying test steps and actions. These modifications help train the machine learning model in the background, aiding in the generation of additional test cases. In summary, the auto generation of code in algoQA is a user-driven process that leverages machine learning to suggest and produce test cases based on specified features, with opportunities for user review and system learning to improve future generations. Success should be measured by outcomes, such as: - Reduction in test authoring time - Faster conversion from inputs to test scenarios - Improved consistency across scenarios It should not be measured by isolated expectations of perfect output. GenAI performs best when: - Input is clear and specific - Relevant context is available and retrievable (Offline model is necessary) - Output is reviewed and refined Treat the generated scenarios as a first draft, not a final artifact. Output quality improves through: - Better context engineering - Improved RAG retrieval strategies - Clearer prompt templates - Higher quality source scenarios These are engineering improvements, not automatic model learning. No, not automatically. Adding scenarios improves the system only if those scenarios are explicitly retrieved and included as context through the RAG mechanism. Simply storing more scenarios does not influence the model's behavior by itself. If you prefer not to record test steps using algoQA, you can use the prompting feature where you can generate test steps in Gherkin language. algoQA generates test cases that can be reused across test scripts. Yes, if an element is healed during execution, the update is reflected in the object repository and can be pushed back to version control (e.g., Git), maintaining synchronization with ongoing CI/CD usage. Impact-based testing is a testing strategy that focuses on identifying and executing test cases that are most likely to be affected by a change in the application. algoQA suggests which regression test cases to run after a change using the following methods: 1. Test Impact Analysis: algoQA performs test impact analysis to determine which test cases are most likely to be affected by a change. This analysis is based on the relationship between the changed component and the existing test cases. 2. Test Prioritization: algoQA prioritizes test cases based on their impact and importance. Test cases that have a higher impact or are more critical are given priority. 3. Test Automation: algoQA's auto test case generation capability generates test cases for the changed feature based on user input. These test cases are executed to ensure the functionality of the changed component. 4. Test Execution: algoQA executes the prioritized test cases to verify the application's functionality after the change. The test results are reported, and any failed test cases are flagged for further investigation. Yes, algoQA features auto-healing scripts for locator/label changes, allowing stable operation across different environments (dev, staging, UAT, prod). Yes, Auto-heal changes are tracked in test reports so users can review any automated script edits, changes, failures, and healing events. This supports audit trails and overnight batch test management. Auto-healing works by detecting locator/label changes during execution. Failed script runs trigger automatic fixes, updating the script repository where possible, with major design changes flagged. The platform supports multiple locator strategies such as parent, child, sibling relationships, custom XPaths, etc. If automated methods cannot resolve the element, manual locator entry is possible. Upon script failure due to locator changes or label modifications, agentic AI in algoQA updates the locator and re-executes the script. Updates are reflected in both the script and object repository, ensuring minimal manual intervention in such cases. Test data is generated based on the scenario automated. Test design and Script execution. algoQA does not use AI to generate scripts. Auto-healing can be turned on or off, either globally or locally for specific features you don't want subjected to it. When it does heal, it's useful for superficial changes like locator or label changes — for example, if a button label changed from "Create" to "New," it treats those as the same element. It will not auto-heal a broken workflow chain, since that kind of change is more likely to reflect a design change or an actual defect rather than a cosmetic one. Yes, When auto-healing proposes a UI-related change, the end user can review it and decide whether it's appropriate, then approve or decline it. Only after approval are the scripts actually updated — so nothing is silently accepted without a human checkpoint. It functions similarly to a static code analyzer, but for feature files. It flags missing assertions and recommends where they should be added, checks for good documentation practices, and suggests improvements when a step doesn't have a meaningful name — essentially surfacing readability and coding-standard issues, similar to how a team would define "good documentation practice" for test authoring. The same scenario executed across multiple datasets is treated as a single scenario rather than counted as duplicates. Beyond that, the engine highlights scenarios that are 95–100% similar as well as those in the 90–95% range, so teams can identify and consolidate near-duplicate test cases instead of maintaining redundant ones. This is aimed at helping optimize the overall test suite. The underlying model for deduplication is not trainable. However, the user can customize the level of similarity for determining duplicates both at feature file and test script levels.

AI & Auto-Healing