SOSIS connects AI-assisted software testing with actionable issue reporting

Software teams often spend considerable effort reviewing test outputs and translating failures into development tasks. As products evolve and the need for testing grows, this work can become increasingly demanding. The ITEA project SOSIS developed a workflow that addresses this challenge by bringing testing and issue reporting closer together.

The integrated testing workflow developed by the project consortium connects test selection, execution and AI-assisted result analysis with automatic issue creation. Combining OTO4T, an AI-assisted testing tool, with DEVASSIST, an AI-assisted development tool, to bridge the gap between identifying a software failure and providing developers with information they can act on.

OTO4T identifies relevant test areas, selects and executes tests, and monitors their results. Test outputs are stored consistently and analysed with AI support to help identify failures and affected areas. Through its integration with DEVASSIST, the workflow supports the automatic creation of issues with clear descriptions, appropriate labels and suitable developer assignments. This can reduce manual log review, report preparation and issue routing, helping failures reach the right developer sooner.

DEVASSIST supports developers in investigating these issues by finding similar past issues and relevant source-code files. This can reduce the time spent searching for context before investigating a failure and can be particularly useful for developers who are new to a project or unfamiliar with its codebase.

With these advancements, the OTO4T-DEVASSIST integration lays the foundation for a more efficient feedback loop between testing and development, connecting test results with the information developers need.

Next steps

To assess the quality of the AI-generated recommendations, the project has introduced an evaluation mechanism that scores their accuracy and consistency. Further work will expand test scenarios, strengthen the OTO4T–DEVASSIST integration and systematically measure recommendation quality, with incorrect or low-confidence results routed for expert review.

Together, OTO4T and DEVASSIST illustrate how SOSIS is translating AI research into practical support for everyday software engineering.

Watch the demonstration video:

More information: https://itea4.org/project/sosis.html