RAG & Search Modules

Use AgentBrains retrieval modules inside Make scenarios to access text and image databases semantically.

Retrieve Text from RAG

Searches an AgentBrains text index and returns relevant text chunks.

Inputs

Output

Returns:

• success (Boolean)
• results[] (Array of results)
  • id
  • score
  • metadata
    • text (The actual context string)
    • name
    • source
    • blobType
    • tags

Use results[].metadata.text as context for LLM prompts.

Using Retrieve Text from RAG as a Make AI Agent tool

1. Add the module as a tool.
2. Select the Index Source.
3. Choose a Custom Index if needed.
4. Allow the AI Agent to fill the Query field.
5. Set Top K to a reasonable value, such as 5.

Retrieve Images from RAG

Searches image assets in AgentBrains and returns relevant image retrieval results.

Inputs

Output

Returns:

• success
• results[]
  • id
  • score
  • metadata

Use this module when an agent needs to find relevant diagrams, product images, screenshots, or visual references.

Recommended RAG workflow patterns

1. Synchronous AI Employee response flow

Use this when AgentBrains or another system sends a customer message to Make and expects a JSON response.

The final webhook response should return:

{
  "message": "The generated reply to send back to the customer."
}

2. RAG answer flow

Use Retrieve Text from RAG when the user asks a broad question and the workflow needs relevant context from the Knowledge Base.

Trigger ➔ AgentBrains: Retrieve Text from RAG ➔ AI Agent/LLM ➔ Output

Use the returned result metadata text as the context for the model.

3. Text + image answer flow

Use Retrieve Text from RAG and Retrieve Images from RAG together when the answer may need both explanation and visual assets.

Trigger ➔ AgentBrains: Retrieve Text ➔ AgentBrains: Retrieve Images ➔ AI Agent/LLM ➔ Output

This is useful for product photos, technical diagrams, screenshots, setup instructions, and visual troubleshooting.

4. Employee-configured agent flow

Use Get an employee configuration when the Make scenario should follow centrally managed AgentBrains Employee settings.

Trigger ➔ AgentBrains: Get Employee Config ➔ AgentBrains: Retrieve Text ➔ AI Agent/LLM ➔ Output

This lets business users update Employee configuration in AgentBrains without editing the Make scenario.