Rishabh Singh is exploring an alternative direction for artificial general intelligence research, focusing on persistent memory, continual learning, and reasoning rather than relying exclusively on larger AI models. As reported by Digital Journal, Singh is the co-founder and chief technology officer of JoinMeNow Inc. and leads Helix Mind, an independent research effort investigating new language model architectures and training methods. The article explains that current AI systems face challenges when retaining knowledge across updates, learning continuously without degrading previous capabilities, and adapting to unfamiliar tasks. It discusses catastrophic forgetting, external memory systems, and the ARC-AGI-2 benchmark, which tests an AI system’s ability to infer new rules from a small number of examples. Singh’s career spans quantitative finance, ultra-low-latency C++ software, credit-risk machine learning, and pricing models. The article also mentions Join-Me-In, Nova Infinity, and Nova Reach. A March 2025 AAAI survey provides context for skepticism about achieving AGI through scaling alone. Helix Mind has not yet published technical papers, code, or benchmark results demonstrating that its proposed approach works.