ConvAI founder says his 2025 open RL decision models anticipated TypeSafe AI’s 2026 non-autoregressive Jev launch

ConvAI Innovations founder Nandakishor Mukkunnoth argues that his 2025 arXiv papers, open weights, dataset, PyPI package, and community posts already demonstrated non-autoregressive schema-based decision models guided by reinforcement learning. He says TypeSafe AI’s 2026 Jev launch presents a similar idea as a breakthrough while offering no papers, open weights, or public training data.

The dispute highlights rising tension over AI attribution as startups and frontier labs converge on faster decision-focused alternatives to text-generating LLMs. If non-autoregressive decision engines gain traction, openness around papers, datasets, and benchmarks may become a key differentiator for credibility and adoption.