Create Prompt with AI
Generate a new prompt from a plain-language description using the Generate with AI feature in the Prompt Workbench.
About
Prompt engineering requires translating a goal into precise instructions a model can follow. Generate with AI removes the initial translation step: you describe what you want the prompt to do, and the platform generates the system instruction and user message for you.
The output lands directly in the Prompt Workbench editor, where you can inspect the structure, edit any part of it, add variables, choose a model, and run it. You stay in full control of the final prompt: the AI gives you a starting point, not a finished product.
When to use
- Unfamiliar task type: You know the outcome you want but are not sure how to structure the prompt instruction for it.
- Getting a first draft fast: A generated draft is a faster starting point than a blank editor, even if you plan to heavily edit it.
- Exploring prompt approaches: Generate multiple versions from different descriptions and compare which structure produces better results.
- Rapid prototyping: When you need something testable quickly and will iterate based on the model’s output.
How to
Navigate to the Prompts Section
From the Future AGI dashboard, locate the navigation panel on the left. Under Build, click Prompts to open the prompts management interface.
Start Generate with AI
In the Prompts section, click Create prompt on the right. In the “Create a new prompt” modal, select Generate with AI (instead of “Start from scratch” or “Start with a template”).
Enter your statement
Describe what you want the prompt to do in a short statement. Be specific about the task, tone, or format you need.
Wait for generation to complete
The platform generates the prompt based on your statement. When it finishes, the prompt editor opens with the generated system and user content filled in.
Review, edit, and run
In the prompt editor you can:
- Rename the prompt and choose a model and parameters if needed.
- Edit the generated system and user text, add variables in
{{brackets}}, or adjust formatting. - Run Prompt to test and see the model’s response.
Optional next steps
After generating your prompt, you can:
- Save it as a template: Reuse the prompt or a tuned version as a template for your team.
- Iterate: Change the statement and regenerate to try different drafts.
Next Steps
Create a Prompt from Scratch
Build a prompt manually with full control over structure and parameters.
Create from Template
Start from a pre-built template for common use cases.
Link Prompts to Traces
Connect prompts to traces to monitor performance in production.
Prompt SDK
Fetch and use prompts programmatically from your application.