https://datatweets.com/courses/ai-agents/planning-and-reasoning/Recent content in Planning and Reasoning on DATATWEETSHugoenCopyright (c) 2025 DatatweetsSun, 28 Jun 2026 09:00:00 +0200https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-1-why-agents-need-to-plan/Fri, 16 Jan 2026 09:00:00 +0200https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-1-why-agents-need-to-plan/A loop that calls tools is not yet a good problem-solver. On multi-step tasks a bare loop can wander. Learn why explicit reasoning helps, and meet the reason-act-observe cycle that underpins every pattern in this module.https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-2-task-decomposition/Fri, 16 Jan 2026 09:00:00 +0200https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-2-task-decomposition/The first planning pattern: decomposition. Ask the model to break a big goal into an ordered list of small steps, then carry them out one at a time, threading each result forward. A plan-then-execute orchestration, verified end to end.https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-3-react-reasoning-and-acting/Fri, 16 Jan 2026 09:00:00 +0200https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-3-react-reasoning-and-acting/The second planning pattern: ReAct. Interleave a reasoning ’thought’ with each action so every tool call is a deliberate choice informed by what came before. Build a ReAct trace on top of your existing agent loop, verified end to end.https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-4-reflection-and-self-correction/Fri, 16 Jan 2026 09:00:00 +0200https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-4-reflection-and-self-correction/The third planning pattern: reflection. After producing a draft, the agent critiques it against the task and revises — looping until the work passes its own check. A reflect-and-revise orchestration, verified end to end.https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-5-guided-project-a-planning-reasoning-atlas/Fri, 16 Jan 2026 09:00:00 +0200https://datatweets.com/courses/ai-agents/planning-and-reasoning/lesson-5-guided-project-a-planning-reasoning-atlas/Put the module together. Atlas plans a constrained trip end to end — decomposing the goal into ordered steps, using ReAct to ground the fact-dependent steps in tool results, then critiquing and revising the draft against the traveler’s constraints. One integrated orchestration, verified end to end.