Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.suppio.ai/llms.txt

Use this file to discover all available pages before exploring further.

Sandbox includes intelligence controls so you can test how different model and reasoning settings affect an agent’s answers before you deploy it.

​
Where to change it

Open your agent in the Suppio dashboard, then select Sandbox from the agent navigation. The intelligence controls appear in the Sandbox sidebar. If you are opening the page directly, use this dashboard URL format: 
/dashboard/workspace/{workspaceId}/agent/{agentId}?tab=sandbox
Replace {workspaceId} with your workspace ID and {agentId} with the agent ID from the agent URL.
1

Open the dashboard

Sign in to Suppio and choose the workspace that owns the agent.
2

Open the agent

Select the agent you want to configure.
3

Open Sandbox

Click Sandbox in the agent navigation.
4

Choose intelligence

In the sidebar, choose Model intelligence and Reasoning effort.
5

Send a test message

Each sandbox message uses the model and reasoning values currently selected in the sidebar.
Sandbox intelligence controls are used for sandbox test runs. They are not a separate saved Intelligence tab.

​
Model options

Suppio shows simple model labels in the dashboard.
Smart is the default model option and is designed for strong answers in most support cases.
Advanced is the highest intelligence option for complex requests.

​
Plan access

PlanModel accessReasoning
FreeSmartMedium
PlusSmartHigh
ProAdvancedHigh
EnterpriseAdvancedHigh

​
Reasoning settings

Reasoning effort controls how much extra work the model spends before answering a sandbox message.
  • Use Low for quick checks and straightforward questions.
  • Use Standard for balanced depth on most questions.
  • Use higher reasoning for complex support questions that need more careful judgment.
  • Model choice and reasoning both affect response quality, latency, and credits.

​
What this affects

The Sandbox intelligence controls affect the test message you send from Sandbox. Use them to compare answer quality, speed, and credit usage while preparing the agent.

​
Tips

  • Start with Smart unless your support questions are consistently complex.
  • Ask the same sandbox question with different intelligence settings to compare results.
  • Watch Usage after testing with Advanced or higher reasoning.
  • If answers are wrong because information is missing, update Context before increasing model strength.