How do I use an AI assistant to learn AI Flowi Workflow?

Give a web-capable AI assistant the AI documentation entry page before describing your goal. Ask for a workflow map covering four stages: trigger, processing, logic or AI, and action. Learn one stage at a time using documented component names and compatible ports, then check the references and test each stage yourself. Use a workflow diagram template to organise that map before asking for component-by-component guidance.

By Ai Flowi · Technical review: Janice Lee, Technical Support, Ai Flowi.

Published · Updated

Prerequisites

  • An AI assistant that can read the linked documentation.
  • A workflow goal you can describe in plain language.
  • Access to AI Flowi Workflow for checking and testing the suggested steps. Use placeholders for keys and private data.

Steps

  1. Share the AI documentation entry page with the assistant. Ask it to read that page before proposing a flow.
  2. Describe the input, the result you want, and any external action. For example: receive webhook JSON, extract named fields, and append them to Google Sheets.
  3. Ask for the complete workflow map first. Request the documented component ID and input/output types for every proposed connection.
  4. Work through the map one stage at a time. Ask what to configure, what sample data to use, and what result to inspect before moving on.
  5. When a step fails, show the first failed component's non-secret error and ask the assistant to check that stage's port types and settings. If the documentation does not establish a capability or field, ask it to say so.

You can use this prompt:

Read the linked AI Flowi Workflow documentation. Show a workflow map for [goal], then teach one stage at a time. Use documented components and extensions, name the ports and their types, and give me a small test after each stage. Say when a component or field is not documented. Do not ask me to paste credentials.

Expected result

You have a map you can compare with the verified tutorials, plus individual tests for the trigger, processing, logic or AI, and action stages. For a first hands-on check, follow the Playground workflow.

Common errors

  • A suggested edge cannot connect: Compare the source output type with the destination's accepted types in workflow basics.
  • The assistant invents a field or component: Ask for its documented ID and port definition; leave the step out if neither is available.
  • A configured flow fails: Inspect the first failing stage and its credentials without sharing secrets with the assistant.

Getting started · First workflow · Common errors