Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.tako.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

Tako Fast Search (POST /v1/knowledge_search) returns results in seconds. It’s optimized for speed — great for autocomplete, inline suggestions, and quick lookups — but has much lower recall than Deep Search. If you need comprehensive data coverage and reasoning over the results, use Tako Deep Search instead.

​
Example

curl -X POST https://tako.com/api/v1/knowledge_search \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "inputs": {
      "text": "Nvidia stock price last 10 years"
    },
    "search_effort": "fast"
  }'

Query: “Top Performing Stock Today”

Try it in our playground
  • Low-latency lookups (< 5 seconds)
  • Autocomplete and inline suggestions
  • Simple factual queries (“Tesla market cap”, “US GDP 2024”)
  • Showing a quick data card alongside other search results

​
Tips

Enhance queries with context: Your code can inject context from the user’s session to improve relevance. For example, if you know the user is in a financial context, send "MSFT stock price last 6 months" instead of just "MSFT". Use the structured response: Each knowledge card includes title, description, embed_url, image_url, webpage_url, and sources. Use these to display interactive charts, static images, or data provenance in your app. Set search_effort explicitly: Pass "fast" to guarantee the fastest response. Omitting it or using "auto" may route to a slower pipeline depending on query complexity.
If you send search_effort: "deep" (or "auto" that resolves to deep) to this endpoint, it will return HTTP 202 with a task_id for async polling instead of a synchronous response. Your client should handle both 200 and 202 status codes. See the Deep Search guide for the polling flow.