Iteris, a new system, utilizes agentic AI and large language models (LLMs) to establish sophisticated research loops for advancing computational mathematics. While agentic AI has demonstrated significant success in general mathematical discovery, such as solving competition problems and tackling research-level conjectures, the specific demands of computational mathematics have received less focused attention. This field uniquely requires not only formal proofs but also extensive numerical experimentation and the construction of adversarial examples. Iteris aims to bridge this methodological gap by implementing autonomous, iterative research cycles that integrate the symbolic reasoning power of LLMs with empirical and analytical feedback. This approach is designed to enhance the ability of AI systems to perform discovery and validation in computationally intensive mathematical domains, moving beyond purely theoretical problem-solving1. This advancement reflects the increasing sophistication of AI in tackling multifaceted scientific challenges. For cybersecurity professionals, it highlights AI’s evolving capacity for complex algorithmic development or cryptographic analysis, potentially streamlining the discovery of vulnerabilities or new defensive mechanisms.