Skip to main content

Agentic Thinking

1 min

What I worked on

Explored how agentic workflows relate to tree-of-thought and chain-of-thought prompting. Tried to trace the lineage of reasoning systems from CoT to ToT and RL.

What I noticed

  • ToT resembles beam search but with reasoning branches
  • CoT and ToT are linked to recursive search and planning
  • Reinforcement learning can implement ToT-like exploration
  • Recursive problem solving traces back decades

”Aha” Moment

na

What still feels messy

Where agentic systems actually diverge from classical RL approaches.

Next step

Map ToT reasoning patterns against beam search and Monte Carlo methods.