Artificial intelligence agents can organise their actions in different ways. A sequential approach follows a defined series of steps; a reactive approach responds to the current situation; and a deliberative approach plans before acting. Each approach can be useful, but each may also be poorly suited to certain tasks or changing circumstances. This paper proposes Adaptive Orchestration as a candidate higher-level approach that selects, combines and revises these strategies according to the requirements of a particular task. The central hypothesis is that an agent able to select its method of operation according to task complexity, available evidence, time, computational resources and changing conditions may perform more effectively than an agent restricted to a single strategy. Adaptive Orchestration is therefore proposed not simply as a fourth execution strategy, but as a possible decision-making layer that governs the selection and coordination of existing strategies. The paper illustrates the four approaches through practical examples, outlines a provisional operational model featuring explicit threshold-based policies, and identifies measurable outcomes through which the hypothesis could be tested via a lean minimal prototype implementation strategy. The proposal remains conceptual. It does not establish that Adaptive Orchestration is novel, that existing agents lack equivalent capabilities, or that a coordinating layer will necessarily improve performance. These questions require strict comparison with established work on agent architectures, planning, reactive control, meta-reasoning and workflow orchestration. The intended contribution is to define a testable research question: whether explicit, context-sensitive selection between execution strategies produces measurable benefits over fixed-strategy alternatives, after accounting for additional computational cost and complexity. This is a conceptual research paper and testable research proposal. Research Status: Initial conceptual proposal. Version 1.
No takes yet. Share an insight, caveat, or question.
Craig Kyrle Strachan Davidson (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: