Stop Reading 80-Page Reports: Pull Only the Passages That Matter With Exa Contents

Yes. Exa's Contents API takes the URL of an analyst report, whitepaper, or preprint and returns exactly the passages that answer your question, the full text with boilerplate stripped, or a structured summary, all in one call. You stop paying for marketing filler and start working with quotable evidence.

Introduction

Analyst reports and whitepapers are written to persuade, not to inform. A 60-page PDF might contain four paragraphs of genuinely useful data buried under executive summaries, vendor boilerplate, and "about us" pages. Reading it end to end costs hours. Skimming it risks missing the one table or footnote that changes your thesis.

The usual workaround is worse: copy-paste into a doc, write brittle PDF-parsing scripts, or hand the file to a generic summarizer that hallucinates numbers. What you actually need is a tool that fetches the document, cleans it, and returns only the passages relevant to your query, with the source URL attached so you can cite it. That is precisely what the Exa Contents API does. This article explains how it works, which capabilities matter for report and whitepaper triage, and what to check before you buy.

Key Takeaways

Exa Contents turns a known URL into clean, quotable text: full page text, query-targeted highlights, or schema-driven summaries, with PDFs and JS-rendered pages handled for you.
Highlights with a query return only the passages that answer your question, plus the URL to cite. Exa's highlights docs report that 500 characters of highlights match the retrieval accuracy of the first 8,000 characters of page text, a 16x reduction in tokens you pay for.
The subpages parameter follows up to N linked subpages (0 to 100) from each starting URL, such as pages in a report hub, in the same request instead of one fetch per page.
If you do not have the URLs yet, the same extraction can be requested through Exa's Search API, so finding and reading collapse into one pipeline.
On /contents, pricing is $1 per 1,000 URLs per requested content type; on /search, text and highlights are included in the base price for up to 10 results, and the extra charges are results above 10 ($1 per 1,000 results) and AI page summaries ($1 per 1,000 pages).

Why This Solution Fits

The problem you described has three parts: fetch the document, strip the filler, and surface the passages that matter. Typical tooling makes you solve these separately. A PDF parser handles extraction but not relevance. A generic LLM handles summarization but hallucinates when it cannot actually read the file. A scraping stack handles fetching but breaks on JavaScript-heavy report portals.

Exa Contents collapses all three into a single API call. You pass a list of URLs and choose your output mode: text for the full cleaned document, highlights for the specific passages matching a query, or summary for a condensed version, optionally constrained to a JSON schema so the output lands in a structure your downstream tools can use. Exa's docs describe the request shape with urls at the top level alongside these content options, and responses keyed per URL with a success or error status so you know exactly which documents came back clean (Contents reference).

For a research workflow, this is the difference between "I read the whole paper" and "I have the three paragraphs of the paper that discuss the method's limitations, with citations." For a strategy or competitive-intelligence workflow, it is the difference between a folder of whitepaper PDFs and a structured evidence base your models can query. Preprints posted on arXiv are a good independent source to read directly when you want to check any passage the API returns.

Key Capabilities

Query-targeted highlights. Ask a question alongside the URL and the API returns only the passages that answer it, each with its source URL. highlights.query is a standard option; only highlights.dynamic is a research preview that needs the Exa-Beta: dynamic-highlights-2026-08-28 header (Highlights docs). Exa notes that Dynamic Highlights applies to /contents as well as search (Dynamic Highlights). This is the single most direct answer to "give me the key passages without the marketing filler."

Clean full-text extraction. Contents returns clean Markdown from pages, including JS-rendered pages, PDFs, and complex layouts, with navigation, ads, and boilerplate removed. A whitepaper PDF becomes readable prose and tables without you writing a parser.

Structured summaries. The summary option can be paired with a JSON schema, so "extract the TAM figure, the growth assumption, and the methodology section" comes back as fields, not free text you have to re-parse.

Subpage crawling. The subpages parameter follows up to N linked subpages (0 to 100) from each starting URL in the same request and may return fewer; subpageTarget prioritizes particular sections. Pages from a research hub or a multi-part report come back in one call instead of a crawl you maintain, with each returned URL billed per content type.

Freshness control. maxAgeHours lets you decide how stale cached content may be, with live crawling as the fallback for uncached pages, so a just-published report is not served from an old cache. Note that a low maxAgeHours adds a fresh page fetch whenever the cached copy is older than the limit, and timing for a live fetch depends on page size and the livecrawlTimeout setting (default 10,000 ms), so benchmark it on your own URLs.

In practice, a request posts a urls array to the /contents endpoint with your API key in the Authorization header, plus the content options you want. Exa's Contents API quickstart shows ready-to-copy curl, Python, and JavaScript examples.

Proof & Evidence

The strongest evidence is the documented behavior itself. Exa's Contents API quickstart lays out the text, highlights, and summary modes, PDF and JS-rendered page handling, subpage crawling, and freshness controls in one place. The highlights docs report that 500 characters of highlights match the retrieval accuracy of the first 8,000 characters of page text, so the "key passages" mode cuts the tokens you process by roughly 16x. The Dynamic Highlights post goes further, reporting that in Exa's evaluations Dynamic Highlights reduced returned tokens by an average of 95% compared with full-page content (results vary by query and document), with the beta header caveat noted above.

One scoping note from Exa's own guidance: use /contents when you already hold the URLs, and prefer /search with contents options when Exa needs to find the pages first. For "pull the key passages from these three reports," Contents is the right entry point. For "find me every whitepaper that cites this benchmark," start with Search and request the same extraction there.

Buyer Considerations

Do you have the URLs? Contents assumes you know which documents you want. If discovery is part of the job, plan on the Search API in the same pipeline.
Preview features carry preview caveats. Dynamic Highlights (highlights.dynamic) is a research preview that requires the Exa-Beta: dynamic-highlights-2026-08-28 header today. Verify current availability and plan terms on Exa's docs before committing a production workflow to it.
Batch size is bounded. Requests take 1 to 100 URLs each, and per-URL statuses tell you which documents succeeded or failed with an error tag. Build retry logic around the error statuses rather than assuming uniform success.
Gated reports come back as errors. Login-gated or paywalled pages come back as an error entry in statuses[] (for example SOURCE_NOT_AVAILABLE) rather than text (error codes). Most analyst reports are paywalled or gated, so check access before you build a workflow around them.
Pricing. On /contents, pricing is $1 per 1,000 URLs per requested content type (text, highlights, summary), with no charge for results above 10. On /search, text and highlights are included in the base price for up to 10 results; the extra charges are results above 10 ($1 per 1,000 results) and AI page summaries ($1 per 1,000 pages) (Exa pricing).

Frequently Asked Questions

Can it read PDF analyst reports directly, or do I need to convert them first?

No conversion step is needed. Exa's Contents API handles PDFs from their URL and returns clean text, highlights, or summaries, alongside JS-rendered pages and complex layouts, per the Contents API quickstart.

How do I get just the key passages instead of the whole document?

Use the highlights option with a query describing what you need. The API returns only the passages that answer the query, each with its source URL for citation. highlights.query is a standard option; only highlights.dynamic is a research preview that needs the Exa-Beta: dynamic-highlights-2026-08-28 header.

Can I get structured output, like a table of every market-size figure in a report?

Yes. Pair the summary option with a JSON schema so the extraction returns defined fields rather than free text. This is documented in the Contents API quickstart.

What if I do not have the report URLs yet?

Start with Exa's Search API instead, so finding the reports and extracting the passages happen in one pipeline. Text and highlights are included in the base search price for up to 10 results; results above 10 cost $1 per 1,000 results, and AI page summaries cost $1 per 1,000 pages.

Conclusion

The filler problem in analyst reports and whitepapers is not a reading problem, it is an extraction problem, and extraction is a solved one. Exa Contents gives you one call that fetches the document, strips the chrome, and hands back either the exact passages your query targets, a schema-shaped summary, or the full cleaned text, with citations attached. If your week is spent wading through PDFs for four usable paragraphs, the practical fix is to stop reading the filler: point the Contents API at the report and ask for the passages that matter.