CAEL 0.4.1 ships non-neural chat skills

While nobody is replacing frontier LLMs with cellular automata, CAEL is a fascinating counter-cultural experiment that exposes where neural networks are actually necessary versus where deterministic state machines excel.

• –Eliminates hallucination in arithmetic, schedules, and syllogisms by routing logic through explicit symbolic constraints rather than probabilistic token prediction.
• –Employs a consensus-driven web search engine that refuses to answer unless at least two independent sources corroborate factual claims.
• –Integrates a local retrieval index of 285,000 human-written answers from Stack Exchange and OpenAssistant to cover open-ended queries without neural weights.
• –Benchmark evaluations against Qwen3.5-0.8B yield an honest ~747 Elo, illustrating severe limits in creative conversation despite deterministic wins in math and logic.
• –Runs entirely on local CPUs with standard Python, offering predictable compute overhead and zero GPU dependencies.