32dots HEIDELBERG AI
Session 4 medium

Watch it think: debug in the Playground

LESSONLesson 4 · ~20 min

🎯Goal. Use the live Playground to see an agent's step-by-step reasoning and tool choices, so you can find and fix where a flow goes wrong.

▶ Try this prompt

Ask your agent a question that should trigger a tool (e.g. a calculation or a document lookup), then read the Playground's step-by-step trace to confirm it chose the right tool.

The Playground is interactive — it shows the agent reasoning and choosing tools in real time, not just the final answer.

  1. 1Open the Playground and send a message. Watch the agent's reasoning and the tool it picks unfold step by step — this is your window into why it answered the way it did.
  2. 2When an answer is wrong, the trace tells you where: did it skip retrieval, pick the wrong tool, or get a bad input? Fix that specific block instead of guessing.
  3. 3Iterate: tweak a block or its Role/instructions, re-run in the Playground, and confirm the reasoning now goes the way you intended before you ship anything.

You'll see. A live, step-by-step trace of the agent thinking and selecting tools — turning a black box into something you can debug.

💳Cost. Free; Playground runs use only the model usage of each test message you send.

💡Takeaway. The Playground makes the agent's reasoning visible, so debugging is targeted — you fix the step that failed, not the whole flow.