Credo: Declarative Control of LLM Pipelines via Beliefs and Policies

2026-04-15

AI Summary

Current agent systems rely on imperative control, transient memory, and logic embedded within prompts, resulting in behaviors that are opaque, brittle, and difficult to verify. This work proposes a declarative control paradigm centered on beliefs and policies: semantic states are modeled as structured beliefs, and declarative policies defined over these beliefs dynamically govern the execution logic of large language model (LLM) pipelines—including model selection, retrieval, and corrective re-execution—without requiring modifications to the underlying code. Built upon a database-backed semantic control plane, this approach substantially enhances system adaptability, auditability, and modularity in decision-making scenarios.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Multiagent Systems: Multiagent Learning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Agentic AI systems are becoming commonplace in domains that require long-lived, stateful decision-making in continuously evolving conditions. As such, correctness depends not only on the output of individual model calls, but also on how to best adapt when incorporating new evidence or revising prior conclusions. However, existing frameworks rely on imperative control loops, ephemeral memory, and prompt-embedded logic, making agent behavior opaque, brittle, and difficult to verify. This paper introduces Credo, which represents semantic state as beliefs and regulates behavior using declarative policies defined over these beliefs. This design supports adaptive, auditable, and composable execution through a database-backed semantic control plane. We showcase these concepts in a decision-control scenario, where beliefs and policies declaratively guide critical execution choices (e.g., model selection, retrieval, corrective re-execution), enabling dynamic behavior without requiring any changes to the underlying pipeline code.

Problem

• agentic AI
• declarative control
• semantic state
• belief-based reasoning
• verifiable behavior

Innovation

• declarative control
• belief-based reasoning
• semantic state
• policy-driven execution
• LLM pipelines