Web Search

Web Search gives your assistant access to current information from the internet, beyond the model’s training data cutoff.

​

How It Works

1. Query generation — the assistant generates search queries from conversation context
2. Web retrieval — searches the web for relevant, current information
3. Synthesis — combines web results with the assistant’s knowledge
4. Response — returns up-to-date information

​

Enabling Web Search

Set web_search="Auto" when sending a message. The assistant decides when web search is helpful. Default is "off".

import requests

response = requests.post(
    "https://app.backboard.io/api/threads/messages",
    headers={"X-API-Key": "YOUR_API_KEY"},
    json={
        "thread_id": thread_id,
        "content": "What are the latest developments in AI this week?",
        "web_search": "Auto",
        "stream": False
    }
)
print(response.json()["content"])

​

Web Search + Memory + Streaming

You can combine web search with memory and streaming in a single message:

response = requests.post(
    "https://app.backboard.io/api/threads/messages",
    headers={"X-API-Key": "YOUR_API_KEY"},
    json={
        "thread_id": thread_id,
        "content": "Based on what you know about my preferences, find me the latest Python framework news",
        "web_search": "Auto",
        "memory": "Auto",
        "stream": True
    },
    stream=True
)
for line in response.iter_lines():
    if line:
        print(line.decode())

​

Options