Start and Run

Once a plan is created, you can trigger it manually or through conversation. In Intric v4, all your plans “live” on the front page under the Plans tab. For even faster access, you can also bookmark specific plans for each assistant, allowing them to appear in the sidebar for easy execution.

This article covers how to start executions and monitor their progress.

How to start a plan

Select the plan you wish to execute from the sidebar or Plans menu.
Upload required files if the plan has “Expected Inputs.”
Start execution.

You can also ask the assistant to run a specific plan via chat. If you upload a file in the chat and then request a plan execution, that file is automatically made available to all steps in that run.

Step execution logic

Each step runs as an isolated task with its own context. Steps execute sequentially—the next step cannot start until the previous one completes.

What each step receives:

The step instruction (the specific prompt for that step).
The assistant’s system prompt and knowledge base.
Access to all tools enabled for the assistant.
Relevant attachments or files.

Accessing previous results

Steps don’t automatically receive output from earlier steps. Instead, the AI fetches specific outputs or tool calls from previous steps as needed. This allows for flexible references—for example, step 5 can specifically pull data generated in step 2.

Plan executions (Monitoring)

Each execution provides real-time feedback on its status:

Overall status: Indicates if the plan is running, completed, or failed.
Step status:
- Green: Successfully completed.
- Blue: Currently in progress.
- Red: Encountered an error or failed.

Viewing execution details

Click Details on any execution to view:

Specific output from each completed step.
Which tools were called during the process.
Technical error information for troubleshooting failures.

Understanding failures

Step-level failures (Red Cross): The AI determined it could not complete that specific task. The plan may still attempt remaining steps depending on the logic.
System-level failures: Indicates a platform issue (such as rate limits or provider outages). These are set programmaticlly rather than by the AI.