Generation Assistant – User Guide
The Generation Assistant allows you to create powerful, no-code AI pipelines tailored to your unique needs. Through an intuitive interface, you can define input types, configure prompt-based actions, and design the desired output — all without writing a single line of code.
Once created, assistants can be tested live, saved as drafts, deployed, and even shared with others for interaction.
Step 1: Define Inputs
Start by defining the Inputs — these are the fields users will fill when interacting with your assistant.
Each input consists of:
- Label: A unique identifier (used for referencing in prompts).
- Description (optional): Helper text for the user.
Supported Input Types:
- Text: Simple free-form input.
- Dropdown: A list of selectable options.
- File Upload: Accepts the following file formats:
.pdf,.docx,.txt,.md,.html,.csv
You can add multiple inputs and organize them as needed.
Step 2: Define Actions (Prompt-Based Logic)
Next, configure the Actions — the core intelligence powered by prompts.
🛠 Prompt Setup:
- Add one or more prompts to define actions.
- Use
{{input_label}}syntax to dynamically reference user inputs. - Additional options:
- Context Files: Upload up to 5 documents to provide extra context to the assistant.
- System Prompt: Define assistant behavior and tone globally (e.g., “You are a concise technical writer.”)
Example Prompt:
You can chain multiple prompts to perform sequential tasks.
Step 3: Define Output
Specify the Output — the expected format of the assistant’s response.
Examples include:
- A short text summary
- A LinkedIn-style post
- Tabular data from CSV/HTML
- Any custom structure based on your needs
You can iterate on the output structure using live previews.
Live Preview & Iteration
Once Inputs, Actions, and Output are defined:
- Click Preview to test the assistant.
- If the output isn't what you expected:
- Modify the input, prompts, or output configuration and re-test.
💾 Saving Behavior:
- Assistants are saved as drafts until explicitly deployed.
- You can revisit drafts any time to make changes.
🚀 Deployment & Sharing
When you're ready:
- Click Deploy to make your assistant live.
- The assistant will then be:
- Available in the workspace where you built it.
- Accessible via:
https://assistant.initializ.ai— for users who’ve been granted access.
Sharing Assistants:
- Share access with specific email addresses.
- Shared users can:
- Use the assistant.
- NOT modify the pipeline (only the creator/admin can edit it).
Summary
| Feature | Description |
|---|---|
| Inputs | Define structured fields (text, dropdown, file) |
| Actions | Chain prompts using tagged input values |
| Context Support | Upload up to 5 docs and a system prompt |
| Output | Specify desired result format |
| Live Preview | Test, refine, and iterate in real-time |
| Drafts & Deploy | drafts; deploy when ready |
| Sharing | Grant access by email; secure interaction-only mode |
Need help creating your first assistant? Contact the team or explore our Assistant Templates to get started faster!