Agentic Problem Frames Paper

Key Takeaways

  • Five canonical agentic problem frames: Search, Planning, Execution, Reflection, Coordination
  • Frame selection determines appropriate architecture — wrong frame → wrong agent design
  • Most enterprise failures map to frame mismatch, not model capability gaps
  • Cognitive depth should match problem complexity: shallow for routine, deep for novel
  • Hybrid frames (Search+Execution, Planning+Reflection) needed for complex real-world tasks
- 五种典型的智能体问题框架:搜索、规划、执行、反思、协同 - 框架的选择决定了恰当的架构——错误的框架会导致错误的智能体设计 - 大多数企业级失败归因于框架错配,而非模型能力不足 - 认知深度应与问题复杂度相匹配:常规任务宜浅,新颖任务宜深 - 复杂的现实任务需要混合框架(如搜索+执行、规划+反思)

Summary

This paper introduces a taxonomy of “agentic problem frames” — a structured way to classify what kind of cognitive work an agent task actually requires. The core argument: most agent system failures are not model failures but architecture failures stemming from applying the wrong problem frame.

本文介绍了“代理型问题框架”的分类体系——一种对代理任务实际所需的认知工作进行分类的结构化方法。核心论点是:大多数代理系统的故障并非模型故障,而是源于应用了错误问题框架的架构故障。

The five canonical frames:

  1. Search: navigating a large information space to find relevant content
  2. Planning: decomposing a goal into a sequence of achievable sub-steps
  3. Execution: carrying out well-specified actions reliably with minimal deviation
  4. Reflection: evaluating outputs against criteria and revising iteratively
  5. Coordination: managing dependencies and communication across multiple agents or humans
五种规范框架:1. **搜索**:在庞大的信息空间中导航以查找相关内容 2. **规划**:将目标分解为一系列可实现的子步骤 3. **执行**:以最小的偏差可靠地执行明确规定的动作 4. **反思**:依据标准评估输出并进行迭代修正 5. **协调**:管理多智能体或人类之间的依赖关系与通信

The paper’s practical contribution is a decision tree for frame selection based on task characteristics: how well-specified is the goal? How large is the action space? How reversible are actions? The decision tree maps to architecture recommendations (single-agent vs. multi-agent, tool-use depth, human checkpoints).

本文的实践贡献在于提出了一个基于任务特征的框架选择决策树:目标定义的明确程度如何?动作空间有多大?动作的可逆性如何?该决策树映射到架构建议(单智能体 vs. 多智能体、工具使用深度、人工检查点)。

Enterprise case study: a contract review workflow initially built as a Planning frame agent (generating review checklists) failed because the real bottleneck was inconsistent Execution. Reframing as Execution + Reflection reduced error rates by 73%.

企业案例研究:一个合同审查工作流最初被构建为规划型智能体(用于生成审查清单),但最终失败,因为真正的瓶颈在于执行不一致。将其重构为执行 + 反思后,错误率降低了 73%。

Relevant Concepts