AWS What's New Scraper - Product Announcements & Launches

Scrape AWS What's New - every Amazon Web Services product announcement, feature launch, and regional expansion - as clean, structured JSON. This actor reads the public feed behind aws.amazon.com/about-aws/whats-new (no login, no API key) and returns headline, canonical URL, publish date, service categories, a short summary, and the full announcement text.

Give it a list of search queries (e.g. bedrock, lambda, s3) and it searches each one, paginates the full 20,000+ announcement archive, merges and de-duplicates the results. Leave the query list empty to pull the latest announcements across all of AWS.

What you get (output fields)

Field	Description
id	Stable announcement id (e.g. launch-4sci4vbcw9jrgasxc68k6e).
headline	The announcement title.
url	Canonical link to the full announcement on aws.amazon.com.
postDateTime	Publish timestamp (ISO 8601, UTC).
dateCreated / dateUpdated	When AWS first created / last edited the entry.
categories	Service / topic tags (e.g. compute, analytics) when present.
summary	First ~300 characters of the announcement body.
bodyText	Full announcement text, HTML stripped (optional - on by default).
bodyHtml	Original HTML body (optional - off by default).
matchedQuery	Which of your queries surfaced this row (null for the latest-feed pass).
directoryId	Source feed id (whats-new-v2).

Input

Field	Type	Default	Description
queries	array of strings	[]	Keywords to search. Empty = latest feed across all AWS.
maxResultsPerQuery	integer	100	Max announcements per query (paginates automatically; up to 5000).
sortOrder	desc / asc	desc	Newest first or oldest first, by publish date.
dateFrom	string	-	Only announcements published on/after this ISO date (YYYY-MM-DD).
dateTo	string	-	Only announcements published on/before this ISO date.
locale	string	en_US	Content locale (e.g. ja_JP, de_DE, fr_FR).
includeBodyText	boolean	true	Include the full plain-text body.
includeBodyHtml	boolean	false	Also include the raw HTML body.

Example input

{
  "queries": ["bedrock", "lambda"],
  "maxResultsPerQuery": 25,
  "sortOrder": "desc",
  "includeBodyText": true,
  "includeBodyHtml": false
}

Example output (one row)

{
  "id": "launch-4sci4vbcw9jrgasxc68k6e",
  "headline": "OpenAI GPT-5.4 and GPT-5.5 models now available in US East (N. Virginia) on Amazon Bedrock",
  "url": "https://aws.amazon.com/about-aws/whats-new/2026/06/openai-gpt-us-east-virginia-amazon/",
  "postDateTime": "2026-06-11T00:12:00Z",
  "dateCreated": "2026-06-11T00:13:22+0000",
  "dateUpdated": "2026-06-11T01:17:23+0000",
  "categories": [],
  "summary": "Today, AWS announces the expanded availability of OpenAI's GPT-5.4 and GPT-5.5 models, which are now available in the US East (N. Virginia) Region on Amazon Bedrock...",
  "matchedQuery": "bedrock",
  "directoryId": "whats-new-v2",
  "bodyText": "Today, AWS announces the expanded availability of OpenAI's GPT-5.4 and GPT-5.5 models..."
}

Use cases

• Competitive & market intelligence - track when AWS ships features in a space you care about (AI/ML, databases, security) and build a dated timeline.
• Cloud consultancies & DevRel - monitor launches by service or keyword to brief clients, write newsletters, or trigger internal alerts.
• Region / availability tracking - watch for "now available in
• Research datasets - pull the full historical archive (20k+ rows) for trend analysis on AWS's release cadence by topic and date.

FAQ

Where does the data come from? The public AWS What's New JSON feed (the same endpoint the AWS website uses). No credentials are needed.

How many results can I get? The archive holds 20,000+ announcements. Set maxResultsPerQuery as high as you need (up to 5000 per query) and the actor paginates for you.

How do I get the whole archive instead of a keyword search? Leave queries empty and raise maxResultsPerQuery; optionally set dateFrom to bound the window.

Why are some categories empty? AWS doesn't tag every announcement. The field is best-effort and populated whenever the feed provides tags.

How is it billed? Pay per result - you're charged once per announcement returned.