Revisable by Design: A Theory of Streaming LLM Agent Execution

2026-04-25

AI Summary

Traditional large language model (LLM) agent execution paradigms force users to trade off between waiting for erroneous outputs and losing progress upon interruption, lacking dynamic intervention capabilities during execution. This work proposes a streaming execution paradigm that models agent execution and user intervention as a concurrent, interleaved bidirectional process. It introduces a reversibility taxonomy—idempotent, reversible, compensatable, and irreversible—to guide interaction design. Drawing on conflict analysis theory, the authors formulate the structurally optimal Earliest-Conflict Rollback rule and a complementary Revision Absorber reactive algorithm to efficiently incorporate user revisions. Experiments on StreamBench demonstrate that this approach maintains output quality while reducing wasted completed work by nearly an order of magnitude compared to full-restart baselines.

Technology Category

• Multiagent Systems: Adversarial Agents
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Humans and AI: Intelligent User Interfaces

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: Agentic search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Current LLM agents operate under an implicit but universal assumption: execution is a transaction -- the user submits a request, the agent works in isolation, and only upon completion does the dialogue resume. This forces users into a binary choice: wait for a potentially incorrect output, or interrupt and lose all progress. We reject this assumption and propose the stream paradigm, in which agent execution and user intervention are concurrent, interleaved processes sharing a bidirectional channel. We formalize this paradigm through a reversibility taxonomy that classifies every agent action as Idempotent, Reversible, Compensable, or Irreversible, and arrive at a core conclusion: an agent's flexibility is bounded by its reversibility. We prove that conflicting compensable actions impose unavoidable adaptation costs and that conflicting irreversible actions make full specification satisfaction impossible -- these costs are properties of the action space, not of the algorithm. Guided by this insight, we present the Revision Absorber, a reactive algorithm based on the Earliest-Conflict Rollback rule that is structurally optimal under mild assumptions. Experiments on StreamBench with real LLM agents validate all predictions: the Absorber matches the quality of a brute-force full-restart baseline while wasting an order of magnitude fewer steps of already-completed work, turning mid-execution revisions from a dead-end into a first-class interaction.

Problem

• LLM agents
• streaming execution
• user intervention
• reversibility
• mid-execution revision

Innovation

• streaming execution
• reversibility taxonomy
• Revision Absorber
• Earliest-Conflict Rollback
• LLM agents