Amazon quietly expanded one of the more practical features in its AgentCore platform this week. Web Search on Amazon Bedrock AgentCore — which went generally available in us-east-1 back in June — now supports new regions including eu-west-1 and ap-northeast-1, along with two new capabilities: domain filtering and published-date filtering.

For enterprise teams building AI agents on AWS, this matters more than a typical regional expansion.

What Changed

The August update brings three changes to AgentCore Web Search:

New regions. EU and APAC coverage means teams subject to GDPR or similar data residency requirements can now run web-grounded agents without routing queries through US infrastructure. This removes one of the key blockers for European enterprise adoption.

Domain filtering. Agents can now be instructed to search only within specific domains or to exclude domains explicitly. An agent supporting a financial services team might pull only from approved regulatory sites, Bloomberg, and Reuters. A legal research agent might exclude social media and consumer news entirely.

Published-date filtering. Agents can now restrict results to content published within a defined window. This matters a lot for use cases like compliance monitoring, competitive intelligence, and news tracking — where returning a 2023 article as “current information” is worse than returning nothing.

Pricing remains at $7 per 1,000 queries, and results are delivered within the AWS network. Web queries never leave the customer’s AWS environment.

Why This Is Not Trivial

Most enterprise AI agents need access to current information that isn’t in their training data. The standard approaches — RAG with a vector database, tool use calling a third-party search API — both have significant drawbacks at enterprise scale.

Third-party search APIs (Bing, Serper, Tavily) route your queries through external infrastructure. For industries with strict data handling requirements — financial services, healthcare, government — that’s often a compliance non-starter.

Building and maintaining your own search infrastructure is the alternative, but it’s expensive, slow, and outside most teams’ core competency.

AgentCore Web Search threads that needle: it’s built on Amazon’s own search infrastructure (the same stack powering Alexa+ and the Kiro developer product), it stays inside your AWS environment, and it’s managed, so there’s nothing to operate. The $7 per 1,000 queries pricing is competitive with third-party options once you factor in compliance overhead.

The domain and date filtering additions push this from “nice to have” to “production-viable” for a wider set of use cases. Before these features, an agent could return a well-sourced but outdated article as if it were current. Now you can enforce temporal relevance at the infrastructure level.

What This Means for Business

If your team is building AI agents on AWS and you’ve been putting off web grounding because of data residency, domain scope, or information freshness concerns, those objections just got smaller.

This is particularly relevant for:

• Financial services teams running compliance monitoring or market research agents, where the source matters as much as the content
• Healthcare organisations that need current clinical information without external API exposure
• European enterprises that struggled to justify us-east-1 only availability against GDPR requirements
• Any team building competitive intelligence or news monitoring agents that need freshness guarantees

The feature also signals something about where AWS is heading with AgentCore more broadly. The platform launched with a focus on orchestration and memory. Web grounding with these controls is more of an enterprise reliability feature — the kind of thing that closes deals, not demos.

For teams evaluating AWS as an AI agent platform, this makes the infrastructure story considerably stronger. The question is no longer whether you can get real-time web access in a compliant way; it’s what you build with it.

---

Enterprise DNA helps business teams design and deploy AI agents that work in production. If you’re evaluating AWS AgentCore or planning your enterprise AI agent architecture, book a discovery call with our team.