How to Use GLChat DeepResearch

GLChat DeepResearch can be accessed in two ways: through the UI or via API.

Method 1: Using the UI

The easiest way to use GLChat DeepResearch is through the GLChat user interface by enabling Deep Research Mode.

Steps:

1. Open GLChat chat interface

2. Click the Tools button in the chat box

3. Select Deep Research to enable the mode

4. Choose your preferred profile:

   • Essentials — Quick research with key information

   • Comprehensive — Detailed research with extensive sources

   • Internal — Research with access to internal documents (Google Drive).

5. Type your question and send

Features:

• Structured research process with progress indicators

• In-line citations for specific facts and claims

• Complete reference list at the end of the response

• Clickable citations linking to original sources

📚 For detailed UI guide: GLChat Deep Research UI Documentation

Method 2: using the API

For programmatic access or integration into other applications, use the GLChat Deep Research API.

API Endpoint

POST /components/deep-research-rag-internal-application:deep-research-rag-pipeline/run

Request Format

Curl

curl --location 'http://localhost:8000/components/deep-research-rag-internal-application:deep-research-rag-pipeline/run' \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-api-key" \
  -d '{
    "inputs": {
      "query": "What is AI?"
    },
    "config": {
      "search_type": "essentials_deep_research"
    }
  }'

Python

JavaScript

Before you can make requests to the GLChat API, you must obtain API Key. You can obtain them by contacting us via hi[at]glair.ai or via our representative for your company.

Available Search Types

Search Type	Profile	Description
essentials_deep_research	Essentials	Fast, cost-effective research
comprehensive_deep_research	Comprehensive	Thorough, detailed research
internal_deep_research	Internal	Research with Google Drive access

Request Parameters

inputs (required):

• query (string) — The research query

config (required):

• search_type (string) — Profile selection (see table above)

Response Format

{
  "response": "Research result with citations...",
  "references": [
    {
      "content": "Source title",
      "metadata": {
        "url": "https://source-url.com",
        "title": "Source Title",
        "source_type": "web"
      }
    }
  ]
}

What GLChat Handles Automatically

Regardless of access method (UI or API), GLChat DeepResearch handles:

• Pre-processing: Context gathering, history, attachments

• Profile resolution: Profile → Backend implementation mapping

• Use case routing: Deep research vs normal generation

• DeepResearcher invocation: With proper configuration

• Post-processing: Citation formatting, reference structuring

• Fallback: Comprehensive → Essentials on quota exceeded