AI workers for Bright Data

What is an AI worker for Bright Data?

An AI worker for Bright Data is an AI agent scoped to one web-research job, such as tracking how competitors describe their pricing or collecting the listings that appeared this week. It reaches the public web through your connected Bright Data account, and the sites it may visit, the datasets it may run and the budget it may spend are all decided before it runs.

What these workers do

Watch how competitors describe themselves

Reads the pages that matter on a handful of sites and reports what changed in the wording, the pricing, or the claims since last time.

Collect search results for the questions that matter

Runs the searches your buyers run and returns who is ranking, so a shift in the landscape is a report rather than a surprise.

Turn a site into something readable

Crawls a documentation set or a knowledge base into clean text a worker can actually reason over, instead of guessing from a homepage.

Pull structured records from listings

Uses a prebuilt extractor to return products, listings or postings as fields rather than as prose, ready to compare week on week.

How an AI worker runs in Bright Data

The reach is the whole public web, so the scope is the product

Collection spends money, and that is its own switch

People data is a deliberate choice, not a default

What web data is actually good for

Bright Data questions, answered.

Can a worker visit any site on the internet?

Only the ones you allow. Because the reach is so broad by nature, this connection is normally scoped to named sites, and the check happens before the request is sent rather than being an instruction the worker is trusted to follow.

How do we stop a worker running up a large bill?

The jobs that cost real money, crawling a site and running the bulk extractors, sit behind their own switch, separate from the permission to read pages. A worker can be given reading without spending, and the requests it does make can be limited to a particular billing zone.

Can it collect people's personal details?

Not unless you grant it. Personal profile and contact fields are withheld by default, and one rule blocks the whole family of people extractors, so this stays a decision you make rather than a capability you inherit.

Does it change anything on the sites it visits?

No. Everything here is reading: pages, search results and datasets. Nothing is posted, submitted or altered anywhere, and the results come back through the same firewall that decided what could be reached.

What AI workers can do in Bright Data

The tools a worker can be granted in Bright Data once you connect it. A kit grants only the ones its job needs, and you can narrow them further on the worker.

Bright Data Fetch Page

Fetch any public web page through Bright Data's unblocking network and get it back as clean markdown (or raw HTML, plain text, or a screenshot). Works on sites that block ordinary requests. Set `render` to true for pages whose content only appears after JavaScript runs. The URL is checked against this token's Bright Data rules BEFORE the request is billed, so a blocked domain costs nothing. Read `accessInfo`: a token can be scoped to specific domains, so a refusal is a rule, not a broken page.

Bright Data Fetch Pages

Fetch up to 20 web pages in one call through Bright Data's unblocking network, returning the same clean content as brightdata_fetch_page for each. Prefer this over repeated single fetches when you already know every URL. Blocked URLs are dropped before they are billed and the count is reported in `accessInfo`, so a short result list may be scope, not failure.

Bright Data Search Engine

Run a real search-engine query (Google, Bing, Yandex or DuckDuckGo) through Bright Data and get parsed organic results with rank, title, URL and snippet. Narrow with `searchType` for news/images/shopping/videos/jobs, `country` and `language` for locale, and `start` to page deeper — the engine's own paging, not a post-filter, so paging gets you genuinely new results. The query is sent to the engine as typed; you cannot pass a raw engine URL. Result URLs are filtered by this token's Bright Data rules and hidden counts appear in `accessInfo`.

Bright Data Crawl Site

Crawl a site from a seed URL and get every discovered page back as markdown in one call. Narrow it before you run it: `depth` limits how far links are followed, `filter`/`excludeFilter` restrict which paths are visited, and `maxPages` caps the result. The crawl runs to completion inside this call — if it does not finish in time it is cancelled and you are told which knob to lower, so start narrow. Every crawled page URL is checked against this token's Bright Data rules.

Bright Data Structured Scrape

Extract structured records from known URLs using one of Bright Data's prebuilt scrapers — product pages, company pages, listings, posts and so on — instead of parsing HTML yourself. Call brightdata_list_datasets first to find the right `datasetId`. Narrow the payload with `customOutputFields` (pipe-separated column names) and `limitPerInput`. Personal-profile and contact columns are separated out and may be masked for this token; read `accessInfo`.

Bright Data List Datasets

List the Bright Data datasets and prebuilt scrapers this account can use, with each one's id, name, row count and scraper category (people, social, ecommerce, business, finance, code, ai-insights, maps-local). Call this before brightdata_scrape_structured or brightdata_search_dataset to find the right id. Datasets this token is not scoped to are omitted and counted in `accessInfo`.

Bright Data Search Dataset

Query one of Bright Data's marketplace datasets and get the matching rows back, without scraping anything live. Pass a `filter` object of field conditions and a `limit`. Use brightdata_list_datasets to find the `datasetId`. Read `accessInfo`: a token can be scoped to particular datasets or scraper categories.

Bright Data Deep Lookup

Describe a set of entities in plain language and get structured rows back, assembled by Bright Data from across the web — use it when you need a list you cannot reach by a single search or scrape. Add `columns` for the attributes each row should carry and `limit` to cap the size. Requires a token with unrestricted Bright Data reach: because the results come from sources the lookup does not name, a token scoped to particular sites is refused rather than handed unscoped data.

Bright Data List Zones

List the Bright Data zones this token may bill requests to, with each zone's name and type. Pass a zone name to the fetch, search or crawl tools to choose where the cost lands; omit it and the account default is used. Zones this token is not allowed to use are omitted and counted in `accessInfo`.

Bright Data Account Status

Check that the Bright Data connection is healthy and, where this token is allowed to see it, the remaining account balance. Worth calling before a large crawl or structured scrape so an exhausted balance surfaces as a clear answer rather than a failed job. A token scoped to particular sites gets the status but not the balance figure.

Bright Data Usage

Read bandwidth and request totals per Bright Data zone, to see what recent collection has consumed. These totals cover every request the account made through each zone, so they are only available to a token with unrestricted Bright Data reach — a site-scoped token is refused rather than shown activity for targets it cannot reach.