Talk Freely, Execute Strictly: Schema-Gated Agentic AI for Flexible and Reproducible Scientific Workflows

Researchers propose a schema-gated orchestration approach to resolve the trade-off between conversational flexibility and deterministic execution in AI-driven scientific workflows. Their analysis of 20 systems reveals no current solution achieves both high flexibility and determinism, but identifies a convergence zone for potential breakthrough architectures.

• →Large language models can translate natural language goals into executable computation but lack the determinism required for scientific workflows.
• →Interviews with 18 experts revealed competing requirements for both conversational flexibility and constrained, deterministic execution.
• →Analysis of 20 existing systems shows an empirical Pareto front where no system achieves both high flexibility and high determinism.
• →Schema-gated orchestration separates conversational authority from execution authority to potentially resolve this trade-off.
• →Multi-model LLM scoring demonstrated substantial inter-model agreement and could replace human expert panels for architectural assessment.