The Association for Computing Machinery (ACM) has announced the launch of CAIS 2026, the inaugural ACM Conference on AI and Agentic Systems, which will take place from May 26 to May 29, 2026, in San Jose, California.
The conference will bring together researchers and practitioners from academia and industry to address challenges related to the engineering of reliable AI systems. As AI technologies evolve from standalone models to complex multi-component deployments, the conference will focus on approaches for designing, optimizing, evaluating, and maintaining AI systems as dependable software artifacts.
CAIS 2026 will emphasize research examining how AI components are composed, orchestrated, and evaluated within end-to-end systems rather than focusing exclusively on individual model capabilities. The program will address system-level challenges such as composition, optimization, verification, and evaluation that arise when deploying AI systems in real-world environments.
A key theme of the conference will examine the distinction between language-model scaffolds used to bootstrap capabilities from current models and durable AI software systems designed for long-term reliability and efficiency. The conference will also highlight evaluation approaches that remain meaningful as AI models continue to improve, including application-specific measures such as latency, correctness, and verifiability.
CAIS 2026 will feature peer-reviewed research across four primary areas: architectural patterns and composition, system optimization and efficiency, engineering and operations of compound AI systems, and evaluation and benchmarking methodologies.
Submissions will undergo an artifact-centric review process, with eligible papers receiving ACM reproducibility badges under ACM's standardized taxonomy, reinforcing a focus on rigorous and reproducible research.
The paper submission deadline is February 27, 2026, and additional conference information is available through the event website.
Click here to read the original press release.
