Menú
RocketLauncher
Únete gratis

AI & Voice Agents

How to Test and Debug AI Conversations in GoHighLevel Agent Studio

GoHighLevel Agent Studio has a built-in conversation tester that lets you run your AI agent and watch each step in real time. The Transition button jumps directly to the active node on the canvas, and the Execution Timeline shows every step with latency, model settings, tool calls, and completion status. Use these two tools together to find exactly where a workflow breaks or performs unexpectedly.

Actualizado el 2 de junio de 2026

5:06

Resumen. Agent Studio's testing tools in GoHighLevel let you simulate a full agent conversation and trace every node that runs. The Execution Timeline breaks down each step chronologically, showing latency, model temperature, tool invocations, retries, and error details including MCP and API failures. The Transition button keeps you oriented on the canvas by jumping to the node currently being executed.

Pasos

  1. Open the tester for your agent

    Click the Test button inside Agent Studio, or select Tester Agent to start the conversation simulator for your chosen agent.

  2. Send a test message and observe the active node

    Type a message and send it. Click the Transition button to jump the canvas view to the node currently processing your input.

  3. Open the Execution Timeline

    Click Execution Timeline to see every step that ran, in order. Each step shows latency, model settings, output, and whether it completed successfully.

  4. Inspect individual steps for errors

    Click on any step in the timeline to expand full details including tool call inputs, outputs, retry attempts, and error messages for failed API or MCP calls.

Tutoriales relacionados

Preguntas frecuentes

Can I test an agent that starts with a tag trigger?

Yes. Apply the tag to a test contact and the agent will start. The conversation tester still shows the full timeline.

What does the Transition button do?

It scrolls the Agent Studio canvas to show the node that is currently active during your test run.

Can I see why a tool call failed in the Execution Timeline?

Yes. Expanding a failed step shows detailed error logs including MCP and API failure messages.

Constrúyelo con gente que ya lo hizo.

Únete a la comunidad gratis: llamadas en vivo cada semana, un curso para empezar y gente que construye en tu idioma.

Únete gratis en Skool