Introducing Seltz Fetch

You can now turn any URL into LLM-ready Markdown with a single call. The new Seltz Fetch endpoint takes up to 20 URLs at once, fetches them concurrently, and hands back clean, readable content with navigation, footers and ads stripped out, ready to drop straight into a prompt.

Fetch is available now as a new endpoint alongside Search, Answer, and Monitor. Where those endpoints answer a question, Fetch gets you the raw material: the actual page content behind a URL you already have, in a shape an LLM can use.

Seltz Fetch ensures that everything on the page is included in the result, including dynamically rendered content.

How it works

Send a list of URLs and get back one result per URL, in the order you asked for them.

from seltz import FetchStatus, Seltz

client = Seltz(api_key="your-api-key")

response = client.fetch(["https://example.com/"])

for result in response.results:

    if result.status == FetchStatus.FETCH_STATUS_OK:

        print(result.markdown)

from seltz import FetchStatus, Seltz

client = Seltz(api_key="your-api-key")

response = client.fetch(["https://example.com/"])

for result in response.results:

    if result.status == FetchStatus.FETCH_STATUS_OK:

        print(result.markdown)

from seltz import FetchStatus, Seltz

client = Seltz(api_key="your-api-key")

response = client.fetch(["https://example.com/"])

for result in response.results:

    if result.status == FetchStatus.FETCH_STATUS_OK:

        print(result.markdown)

That's a complete request: currently, Fetch requires the pro tier, so tier resolves to "pro" when omitted. Because Seltz fetches URLs concurrently, a batch of 20 takes about as long as its slowest page, not the sum of all of them.

A page that Seltz can’t fetch doesn't fail the call. Instead, it comes back inside the same 200 response with status = "error" and an error.code, so one bad URL never blocks the rest of the batch:

Use cases for Fetch

Grounding an LLM in a specific page. You already have the URL for a document, an article, or a competitor's pricing page, and you just need its content in a clean, promptable format without writing your own scraper and boilerplate-stripping logic.

Following up on links found elsewhere. Pair Fetch with Seltz Search or Answer: use those to find relevant URLs, then Fetch the full pages behind the best results for deeper context than a snippet gives you.

Batch content collection. Fetch up to 20 URLs in one call, such as a list of sources for a research task, a set of competitor pages, or a batch of articles, and get every result back together, correlated by URL.

Try Seltz Fetch

Seltz Fetch is live now. Grab an API key at console.seltz.ai/api-keys and read the Fetch quickstart and API reference to get started.

Ciao, The Seltz Team