AI4SWEng: Strong Foundations, Shared Vision, and Momentum for What Comes Next

At AI4Sec, in our role as the Project Coordinator, we are immensely proud to mark the completion of the first year of AI4SWEng. This milestone stands as a testament to the collective progress achieved and the strategic roadmap we have established for the future. From the project’s inception, our leadership has been driven by the conviction that while software engineering is the heartbeat of modern society, its escalating complexity necessitates a shift toward smarter, AI-driven methodologies. As coordinators, we have focused on ensuring that Year 1 successfully built the structural and collaborative foundations needed to transform this vision into a tangible reality for the software engineering community.

The first year was intentionally dedicated to clarity and alignment. We employed Github Projects for interactively defining requirements, use case specifications, and task details. All project requirements, functional and non-functional, technical, and organizational, have been clearly identified, discussed, and agreed across the consortium. This shared understanding is essential for a project of this scope and ambition and ensures that all partners are working toward the same objectives, using a common language and shared expectations. 

In parallel, we carried out a comprehensive baselining exercise. Rather than moving too quickly into implementation, the consortium invested time in understanding the current state of tools, processes, data availability, and system performance. These baselines now provide a solid reference point, enabling us to measure progress, improvements, and impact in a transparent and meaningful way throughout the project’s lifetime. 

A major achievement of Year 1 is the screening and refinement of KPIs and success criteria, structured around clearly identified KIO groups. The KIO framework helps translate the project vision into concrete, measurable outcomes, while also clarifying responsibilities and task distribution across work packages and partners. Each task now clearly contributes to one or more KIOs, strengthening coherence, accountability, and traceability from research activities to expected impact. 

From a technical perspective, the consortium has converged on a well-defined high-level system architecture. While the detailed architecture will continue to evolve as implementation progresses, the overall architectural vision is now clear and shared. Core components, interfaces, and data flows have been identified, providing a stable backbone for parallel development and future integration. This high-level architecture already reflects key non-functional requirements such as scalability, security, interoperability, and maintainability, ensuring that early design decisions support long-term sustainability. Supporting this architecture is a domain knowledge base that would guide LLMs to make informed decisions based on established ground truths in the relevant Use Cases. This knowledge base is comprised of a pre-defined ontology of the system outlining interactions between levels and sub-levels of the system complemented by individual instances of relevant data specific to the Use Case. The UC-specific Ontology is generated following an Ontology-based Requirements and Specifications Engineering study which synthesises the details of KIO-related Requirements and UC-related Requirements and Specifications. This would allow AI4SwEng Engineering Suite to extract requirements and bounds for code generation, storing database actor information for further inference baseline for the LLM, also understanding and interpreting User Query through domain knowledge available.

Another important aspect of Year 1 has been active risk monitoring and adaptation. The AI and software engineering landscape is evolving at an unprecedented pace, with continuous breakthroughs emerging from industry as well as from academia worldwide. In response, the consortium has regularly revisited and updated its risk analysis. Technical, methodological, and external risks have been refined to account for rapid changes in foundation models, tooling ecosystems, and best practices. This proactive approach allows AI4SWEng to remain relevant, resilient, and well-positioned despite a fast-moving technological environment.

Beyond the technical and methodological achievements, Year 1 has also been about building trust and momentum within the consortium. Communication channels, coordination mechanisms, and decision-making processes are now well established and working effectively. This strong collaborative culture is a key enabler as we move into more implementation-intensive phases.

In summary, the first year of AI4SWEng has delivered what it set out to deliver: first shot to clear requirements, solid baselines, a structured KPI and KIO framework, very high software architecture specification, leading to well-defined high-level architecture, and an updated, forward-looking risk strategy. With these foundations in place, the project enters the next phase with confidence, energy, and excitement, ready to turn vision into concrete, AI-enabled advances in software engineering.