Skip to main content
POST
/
discover
/
sync
curl "https://api.brightdata.com/discover/sync" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <token>" \
  -d '{
    "query": "artificial intelligence trends",
    "filter_keywords": ["Product Manager", "Roadmap"],
    "num_results": 10,
    "format": "json",
    "intent": "latest AI technology developments",
    "include_images": true
  }'

{
  "status": "done",
  "duration_seconds": 10,
  "results": [
    {
      "link": "https://www.trigyn.com/insights/ai-trends-2026-new-era-ai-advancements-and-breakthroughs",
      "title": "AI Trends in 2026: Advancements and Breakthroughs Ahead",
      "description": "5 days ago — Explore top AI trends for 2026, from enterprise adoption to autonomous systems and breakthrough business innovations.",
      "relevance_score": 0.98184747,
      "content": null
    },
    {
      "link": "https://sisuadigital.com/blog/artificial-intelligence-trends-2026-sisua-digital/",
      "title": "10 Artificial Intelligence (AI) Trends That Will Define 2026",
      "description": "Dec 26, 2025 — Discover the 10 Artificial Intelligence (AI) Trends That Will Define 2026: governance, human collaboration, continuous learning, and more.",
      "relevance_score": 0.97948,
      "content": null
    }
  ]
}
The Bright Data Discover API synchronous endpoint returns ranked search results in the same HTTP response, so you do not create a task or poll for results. Use the synchronous endpoint when a search completes inside the 60-second timeout budget and you need results inline, for example in an AI agent, a RAG pipeline or a chat backend. The synchronous endpoint accepts the same request body as the asynchronous Discover API endpoint. Only the response differs: the synchronous endpoint returns the final results payload instead of a task_id.

​
Body Parameters

​
query
string
required
The search query.
Maximum length: 1,500 characters.
​
intent
string
Describes the specific goal of the search to help the AI evaluate and rank result relevance. If not provided, the query string is used as the intent.
Maximum length: 3,000 characters.
For best results, use the following formula:
[PERSONA/CONTEXT]: I am [persona] looking for [use case].
[INCLUDE]: Prioritize [document type 1], [document type 2],
           and [source authority] published [recency if relevant].
[DEPTH]: Focus on [technical level / document section].
[EXCLUDE]: Strictly exclude [noise type 1], [noise type 2],
           and [source type to avoid].
​
mode
string
default:"standard"
required
Controls the search depth and ranking behavior of the request.
  • standard - Provides standard search depth and AI ranking. Best for general use cases where a balance of relevance and speed is needed.
  • zeroRanking - Bypasses AI ranking to maximize the raw volume of results returned. Best when you want to collect as much broad data as possible without relevance filtering.
    In zeroRanking mode, the num_results parameter has no effect on the number of results returned. Additionally, include_content is not supported in this mode.
  • deep - Performs a more exhaustive, broader search. Best when you need comprehensive coverage of a topic and prioritize thoroughness over speed.
  • fast - Optimizes the request for quicker response times. Best for time-sensitive tasks where getting immediate results is the top priority.
​
filter_keywords
array of strings
A list of exact keywords that must appear in the search results. The API automatically applies intext: operators to guarantee keyword inclusion.Example: ["Product Manager", "Roadmap"]
​
format
string
default:"json"
The response format.Available options: json, md
When set to md and include_content is true, the content field returns parsed Markdown instead of raw HTML.
​
include_content
boolean
default:"false"
If true, the response will include the page content in markdown format.PDF support: When a search result links to a PDF file, the API will automatically extract and parse the PDF content. Limitations:
  • Maximum PDF file size: 50 MB
  • Maximum PDF parsing time: 30 seconds
  • If either limit is exceeded, the content field returns empty.
​
include_images
boolean
default:"false"
If true, the response will extract and include an array of images.
​
language
string
default:"en"
The language to search in and return data for.Supported across 31 languages to align with Voyage AI and SERP capabilities, including but not limited to: en (English), es (Spanish), fr (French), de (German), zh (Chinese), ja (Japanese), ar (Arabic), he (Hebrew), ko (Korean), hi (Hindi), pt (Portuguese), and ru (Russian).
​
num_results
integer
The exact number of search results to return in the response.
​
remove_duplicates
boolean
default:"true"
If true, duplicate results will be removed from the response.
​
start_date
string
Search only for content updated from the date specified (format: YYYY-MM-DD).
​
end_date
string
Search only for content updated until the date specified (format: YYYY-MM-DD).
​
country
string
default:"US"
Get search results from a specific country. This accepts all standard 2-letter ISO country codes (e.g., US, GB, DE, IL, FR, JP). This will prioritize content from the selected country in the search results.
​
city
string
Get search results localized to a specific city using SERP uule encoding (e.g., "New York", "Berlin", "Tel Aviv"). For best results, use this in conjunction with the corresponding country parameter.

​
Synchronous vs. Asynchronous

The synchronous and asynchronous Discover API endpoints take identical request bodies. They differ only in how you receive results.
Synchronous (POST /discover/sync)Asynchronous (POST /discover)
ResponseFinal ranked results in the same HTTP responseA task_id to fetch later
PollingNonePoll Retrieve Results until status is done
Round trips12 or more
Timeout budget60 seconds per requestNot bound to a single request
Best forAI agents, RAG pipelines and chat backends that need results inlineLarge or slow searches, batch jobs and background collection
Use the synchronous endpoint when the result feeds directly into an AI integration. An agent, a retrieval-augmented generation (RAG) pipeline or a chat backend can send one request and use the ranked results in the same turn, with no task tracking and no polling loop to build or maintain. Use the asynchronous endpoint when a search may exceed the 60-second synchronous timeout, or when you collect results in the background and fetch them later with Retrieve Results. 
curl "https://api.brightdata.com/discover/sync" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <token>" \
  -d '{
    "query": "artificial intelligence trends",
    "filter_keywords": ["Product Manager", "Roadmap"],
    "num_results": 10,
    "format": "json",
    "intent": "latest AI technology developments",
    "include_images": true
  }'

{
  "status": "done",
  "duration_seconds": 10,
  "results": [
    {
      "link": "https://www.trigyn.com/insights/ai-trends-2026-new-era-ai-advancements-and-breakthroughs",
      "title": "AI Trends in 2026: Advancements and Breakthroughs Ahead",
      "description": "5 days ago — Explore top AI trends for 2026, from enterprise adoption to autonomous systems and breakthrough business innovations.",
      "relevance_score": 0.98184747,
      "content": null
    },
    {
      "link": "https://sisuadigital.com/blog/artificial-intelligence-trends-2026-sisua-digital/",
      "title": "10 Artificial Intelligence (AI) Trends That Will Define 2026",
      "description": "Dec 26, 2025 — Discover the 10 Artificial Intelligence (AI) Trends That Will Define 2026: governance, human collaboration, continuous learning, and more.",
      "relevance_score": 0.97948,
      "content": null
    }
  ]
}
The synchronous response uses the same schema as Retrieve Results. The status field is always done because the results are final when the response returns.
​
status
string
The status of the request. For the synchronous endpoint this is always done.
​
duration_seconds
integer
The time taken to process the request in seconds.
​
results
object[]
A list of sorted search results.

​
Errors

The synchronous endpoint carries over the validation and authentication errors from the asynchronous Discover API endpoint, and adds a 504 error when a search does not finish within the 60-second timeout budget.
StatusError MessageTriggerResolution
400{"error":"Missing query"}query is missing, empty, or nullProvide a non-empty string in the query field.
400{"error":"Invalid query. Max length is 1500 chars"}query exceeds 1,500 characters, or is a non-string type (number, array)Provide a string of 1,500 characters or fewer.
400{"error":"Invalid intent. Max length is 3000 chars"}intent exceeds 3,000 characters, or is a non-string typeProvide a string of 3,000 characters or fewer, or omit to use query as the intent.
400{"error":"Unsupported format. Only \"md\" and \"json\" are supported"}format is not “json” or “md” (e.g. “markdown”, “xml”, “csv”, empty string, or non-string type)Set format to “json” or “md”.
400{"error":"Invalid num_results. Must be a number between 1 and 20"}num_results is 0, negative, greater than 20, or a non-number typeProvide an integer between 1 and 20.
400{"error":"Invalid filter_keywords. Must be an array of strings"}filter_keywords is not an array (e.g. a string)Provide an array of strings, e.g. [“keyword1”, “keyword2”].
400{"error":"Invalid start_date"}start_date is not in YYYY-MM-DD format, is a non-string type, is in the future, or start_date > end_dateProvide a valid past or present date in YYYY-MM-DD format.
400{"error":"Invalid end_date"}end_date is not in YYYY-MM-DD format or is a non-string typeProvide a valid date in YYYY-MM-DD format.
400{"error":"Unsupported country"}country is a string but not a valid ISO 3166-1 alpha-2 codeUse a valid 2-letter country code (e.g. “US”, “IL”, “DE”).
400{"error":"Unexpected fields: <field_names>"}Request body contains unrecognized field namesRemove unexpected fields. Only use documented parameters.
401Credentials are missingNo Authorization header providedAdd Authorization: Bearer <token> header.
401Invalid credentialsToken in Authorization header is invalidProvide a valid API token.
401Auth method is not supportedAuthorization header is missing the Bearer prefixUse the format Authorization: Bearer <token>.
403ForbiddenDiscover API is not enabled for your accountContact your account manager to enable the Discover API.
429Too Many RequestsRate or concurrency limit exceededSlow down requests or upgrade your plan.
500Internal Server ErrorServer-side issueRetry after a few seconds.
504{"error":"Request timed out"}The search did not complete within the 60-second synchronous timeout budgetNarrow the query, reduce num_results, or use the asynchronous Discover API endpoint and poll Retrieve Results.

​
FAQs

When a search does not complete within the 60-second timeout budget, the synchronous endpoint returns a 504 error and no results. Retry with a narrower query or a smaller num_results, or send the same request body to the asynchronous Discover API endpoint and fetch the results later with Retrieve Results.
No. The synchronous endpoint does not return partial results. A request either returns the complete ranked results with a 200 status, or it returns a 504 error with no results.
Yes. The synchronous endpoint returns the same payload schema as Retrieve Results: a status field, a duration_seconds field and a results array. The only difference is that status is always done, because the results are final when the response returns.
Use the synchronous endpoint when the result feeds directly into an AI integration such as an agent, a RAG pipeline or a chat backend, and the search completes within 60 seconds. Use the asynchronous endpoint for large or slow searches, batch jobs and background collection.
Access to the Bright Data Discover API is restricted. It must be manually enabled for your account by your account manager.