SANER 2027 - Agentic AI4SE Track

Call for Papers

Call for Papers

The Agentic AI4SE Track of the 34th edition of the IEEE International Conference on Software Analysis, Evolution, and Reengineering (SANER 2027) invites high-quality submissions on agentic AI systems for software engineering, with a particular focus on software analysis, evolution, maintenance, and reengineering.

This newly launched track provides a forum for researchers and practitioners studying autonomous and semi-autonomous AI agents that can plan, act, use tools, interact with developers, and adapt across realistic software engineering tasks. We welcome work that advances the foundations, design, evaluation, deployment, and governance of agentic AI4SE systems, as well as evidence about their benefits, limitations, and risks.

Submissions should clearly explain how the proposed ideas, systems, or studies address a software engineering problem and why they are relevant to the SANER community. We particularly welcome submissions that study agents as software engineering systems rather than isolated model calls: systems that reason over artifacts, maintain state, invoke tools, coordinate with humans or other agents, and operate under constraints such as correctness, cost, safety, privacy, and maintainability.

Topics of Interest

Topics of interest include, but are not limited to:

Agentic AI systems for software analysis, comprehension, evolution, maintenance, and reengineering;
Autonomous or semi-autonomous agents for bug fixing, program repair, refactoring, migration, modernization, and technical debt management;
Multi-agent workflows for software engineering, including coordination, delegation, negotiation, and conflict resolution;
Planning, memory, tool use, retrieval, execution, and feedback mechanisms for AI agents working with software artifacts;
Agentic approaches for code review, testing, debugging, vulnerability analysis, release engineering, continuous integration, and DevOps;
Benchmarking and evaluation of agentic AI4SE systems, including task design, metrics, reproducibility, reliability, cost, and longitudinal evaluation;
Human-agent collaboration, developer experience, trust, control, explainability, and socio-technical aspects of agentic software engineering;
Safety, security, privacy, accountability, and governance for AI agents that modify, execute, or deploy software;
Industrial experiences, case studies, and lessons learned from deploying agentic AI4SE systems in real software projects;
Infrastructure, datasets, environments, and tools for rigorous research on agentic AI for software engineering;
Software engineering for agentic AI systems, including requirements, architecture, testing, monitoring, maintenance, and evolution of agent-based software systems.

Evaluation Criteria

Submissions will be reviewed by at least three members of the Program Committee of the Agentic AI4SE Track. They will be evaluated based on the following criteria:

Relevance: the extent to which the submission addresses agentic AI for software engineering and is relevant to software analysis, evolution, maintenance, or reengineering;
Originality and novelty: the extent to which the contribution is new and clearly positioned with respect to prior work;
Significance: the importance of the problem and the potential impact of the contribution on research or practice;
Soundness: the extent to which the claims are supported by appropriate methods, evidence, and analysis;
Evaluation quality: the suitability of tasks, baselines, metrics, datasets, user studies, case studies, or other forms of validation;
Open science and verifiability: the extent to which the paper provides sufficient information and artifacts to support independent assessment, replication, or reuse;
Presentation: clarity, organization, quality of writing, readability of figures and tables, and adherence to the submission instructions.

Submission Instructions

Submitted papers must have been neither previously accepted for publication nor concurrently submitted for review in another journal, book, conference, or workshop. All submissions must be written in English and submitted as a PDF.

Submissions should not exceed 10 pages, with up to 2 additional pages for references only, and must conform to the IEEE Conference Proceedings Formatting Guidelines. LaTeX users must use \documentclass[10pt,conference]{IEEEtran} without including the compsoc or compsocconf option. Papers must comply with the IEEE Policy on Authorship, including guidelines on the use of generative AI tools and technologies.

All papers must be submitted electronically through the SANER 2027 EasyChair submission site: https://easychair.org/conferences/?conf=saner27tracks. After clicking “Make a New Submission,” authors will see a list of available tracks. Be sure to select the correct track, i.e., Agentic AI4SE Track.

Important note: the SANER 2027 Agentic AI4SE Track follows a double-anonymous review process. Submissions that do not comply with the double-anonymous policy will be desk-rejected. Author names and affiliations must be omitted; references to authors’ own related work must be written in the third person; acknowledgments and other identifying information must be omitted; and links to code, data, tools, videos, or supplementary material should be anonymized whenever possible.

Accepted Papers

All accepted papers will be included in the IEEE Digital Library as part of the SANER 2027 conference proceedings, provided that at least one author registers for SANER 2027 (full registration, not student registration) and presents the paper at the conference. Papers that are not presented will be removed from the proceedings.

The authors of accepted papers are encouraged to incorporate reviewers’ comments in the camera-ready version. After acceptance, the list of authors cannot be changed under any circumstances, and the author list in the camera-ready version must be identical to that of the submitted paper.