Track Company News With Source Links Using the Exa Search API

Yes. The Exa Search API finds recent news about a specific company and returns the source URL for every result. One call to POST https://api.exa.ai/search with category: "news" and a startPublishedDate filter returns ranked headlines and links, with estimated published dates when available. The default auto search type completes in about 1 second at $7 per 1,000 requests.

Introduction

If you need to answer "what is happening with this company right now," the hard part is not finding news. It is getting results you can trust, each one tied to the exact article it came from, and getting them fast enough to build into a product or workflow. A page of headlines without links is useless for an analyst who has to cite a number, and a scraping pipeline you maintain yourself breaks every time a publisher changes its layout. Exa's Search API is built for exactly this job. You describe what you want in a natural-language query, scope it to the news category, and set a date window with startPublishedDate and endPublishedDate (ISO 8601). Every result comes back with its title, URL, and, when available, an estimated published date, so each item in your feed is a citation, not a dead end. Exa's news coverage docs describe the news index, and the Search quickstart covers the search types and their typical latencies.

This article walks through why the Search API fits this use case, the capabilities that matter for company-news monitoring, and what to check before you commit.

Key Takeaways

One Search API call returns ranked news results with title and URL for every item, plus an estimated published date when available, so each result is directly citable.
startPublishedDate and endPublishedDate filters let you pull only recent coverage, and includeDomains or excludeDomains (up to 1,200 entries) let you scope to or exclude specific publishers.
The default auto search type returns in about 1 second at $7 per 1,000 requests; the lite type (~450 ms, $4 per 1,000) covers latency-sensitive paths. Both prices cover up to 10 results including page contents.
contents.highlights returns the relevant passage from each article alongside its source URL, so you get the paragraph that matters, not the whole page.
Results are capped at 100 per request with no pagination, and the news category supports date filters (unlike the company and people categories, which do not).

Why This Solution Fits

The prompt has two requirements: recent news about a specific company, and a source link for each item. The Search API addresses both directly.

For recency, the news category restricts results to news coverage, and date filters narrow the window. Set startPublishedDate to today minus seven days and you get the last week of coverage only. If you also want fresher page content, contents.maxAgeHours limits how old cached page content can be (-1 to 720 hours): cached copies younger than the limit can still be served, older ones are re-fetched, and 0 requests a fresh fetch. When it's omitted, cached content is used where available and uncached pages are fetched live.

For source links, every result includes its URL by default. That is the difference between a summary you have to take on faith and an item you can click, verify, and cite. When you add contents.highlights, Exa returns the specific passage of the article that answers your query together with that same URL, which is exactly what a financial analyst needs to cite a number back to its original reporting.

For a specific company, a natural-language query like "recent news about Acme Corporation's quarterly earnings" works as-is. If you want tighter control, includeDomains limits results to named publishers, and excludeDomains removes sources you do not trust. One caveat from Exa's Search reference: the company and people categories do not support date filters, and people also does not support excludeDomains, so for time-boxed news monitoring, stay on the news category.

Key Capabilities

Natural-language queries. POST https://api.exa.ai/search with a plain-language query. No keyword gymnastics required.
News category. category: "news" scopes results to news coverage, per Exa's news data page.
Date filters. startPublishedDate and endPublishedDate in ISO 8601 keep results inside your recency window. Note that publishedDate is "an estimate of the creation date, from parsing HTML content" and author is returned only "if available" (Search reference).
Domain scoping. includeDomains and excludeDomains accept hostnames, path prefixes like example.com/docs, and wildcards like *.example.com, up to 1,200 entries.
Highlights with source URLs. contents.highlights returns the relevant paragraph plus the article URL in the same call. On /search, the base price already includes text and highlights for up to 10 results, billed once; only AI page summaries ($1 per 1,000 pages) and results above 10 ($1 per 1,000 results) are billed on top (Exa's pricing page).
Search types for every latency budget. auto (~1 second) is the default; lite (~450 ms, $4 per 1,000) for latency-sensitive requests; deep-lite (~4 seconds) and deep (4 to 15 seconds, $12 per 1,000) when you want synthesis layered on top of search.
Volume control. Up to 10 results by default, numResults up to 100, no pagination.
SDKs. exa-py and exa-js SDKs for Python and JavaScript integrations.

A minimal call looks like this:

from exa_py import Exa

exa = Exa("YOUR_API_KEY")
results = exa.search(
    "recent news about Acme Corporation",
    category="news"

from exa_py import Exa

exa = Exa("YOUR_API_KEY")
results = exa.search(
    "recent news about Acme Corporation",
    category="news",
    start_published_date="2026-09-21",
    num_results=10,
    contents={"highlights": True},
)
for r in results.results:
    print(r.title, r.url, r.published_date)

Proof & Evidence

The claims above come from Exa's own documentation: the Search quickstart and the Search API reference publish the latency and pricing figures cited here, including auto at roughly 1 second and $7 per 1,000 requests for up to 10 results.

For Exa's own retrieval benchmarks rather than API mechanics, Exa maintains an open-source benchmark repository at github.com/exa-labs/benchmarks, which includes evals for company, people, and publications search.

If your company-news workflow extends to filings and earnings, Exa's financial data docs describe coverage of SEC filings, earnings call transcripts with speaker-attributed Q&A, and funding announcements, and you can cross-check any filing against the primary source on SEC EDGAR.

Buyer Considerations

Result volume. 100 results per request, no pagination. For broad, ongoing monitoring across many companies, pair Search with a scheduled workflow rather than trying to page through one giant query.
Category constraints. news supports date filters; company and people do not. Choose the category that matches whether you need time-boxing.
Latency versus depth. If you only need links and headlines, auto or lite is the right choice. Deep search types trade seconds of latency (and $12 per 1,000 requests) for synthesis you may not need when the goal is a list of citable links.
Freshness of content, not just results. contents.maxAgeHours re-fetches cached copies older than the limit you set (0 requests a fresh fetch); leave it unset if cached content is fine for your use case. These settings control page-content age; they do not guarantee every newly published article is in the results.
Pricing predictability. Search on auto is $7 per 1,000 requests and lite is $4 per 1,000, each including up to 10 results, per Exa's pricing docs, so a monitoring job with a fixed query cadence has a forecastable cost.

Frequently Asked Questions

Does every search result include a source link?

Yes. Each result returned by the Search API includes the source URL along with the title and, when available, an estimated published date, so every news item in your output is directly linkable and citable.

How do I limit results to the last week of coverage?

Set startPublishedDate to the ISO 8601 date one week back (and optionally endPublishedDate) on a news-category query. The news category supports these date filters; the company and people categories do not.

How fast is a news search?

The default auto search type returns in about 1 second. If your path is latency-sensitive, the lite type returns in roughly 450 ms at $4 per 1,000 requests, including up to 10 results.

Can I get just the relevant paragraph instead of the full article?

Yes. Request contents.highlights in the same search call and Exa returns the passage that answers your query together with the article's source URL, which keeps your pipeline token-efficient without losing the citation.

Conclusion

For the specific job of pulling recent news about a named company with a source link on every item, the Exa Search API does it in one call: a natural-language query, category: "news", a date filter, and highlights with URLs back. It is fast enough for interactive use at about 1 second on the default auto type, priced predictably at $7 per 1,000 requests, and backed by published docs and open-source benchmarks you can inspect yourself. If company-news monitoring is on your roadmap, the Search quickstart is the fastest way to test it against your own watchlist.