AWS AI Agents Get Two New Services for Business Context and Security
Amazon just gave AWS AI agents two new tools. The company says AI agents have two big weak spots right now. First, they do not know enough about a company’s own business. Second, they can move too fast for normal security to keep up. So AWS built two new services to fix both problems. The news came at the AWS Summit in New York on June 21, 2026.
First, here are the plain words. AWS (Amazon Web Services) is Amazon’s cloud business. “Cloud” means you rent computers and software over the internet. You do not buy your own. An “AI agent” is an AI program that can do tasks by itself. It does not just chat with you. It can read data, run checks, and take actions. That is useful. But it can also be risky. The agent might get things wrong. Or it might move too fast.
The Two Problems AWS Wants to Fix
AWS named two clear problems. One is “business context.” The other is security.
Business context means knowing the facts and rules of one company. For example, which customer is which. Or what a product code means. Or who is allowed to see what. Most AI agents do not know these things. So they guess. And sometimes they make up answers. People call these made-up answers “hallucinations.”
The second problem is speed and safety. AI agents act very fast. AWS says they can move faster than old security tools can catch. That makes it harder to find and fix weak spots in code. “Code” is the set of instructions that runs an app or a website.
AWS Context: Giving Agents Business Knowledge
The first new service is called AWS Context. It works like a shared knowledge base for AI agents. A “knowledge base” is a smart store of facts. The agent can look things up there before it acts.
AWS Context builds a knowledge graph from a company’s own data. A “knowledge graph” is a map that shows how things connect. For example, it shows which customer links to which order. AWS Context pulls this from databases, documents, emails, and chats. It can also read from the AWS Glue Data Catalog and from outside sources.
The goal is simple. When an agent knows real business facts, it stops guessing. So it makes fewer mistakes and gives more correct answers. AWS Context also has built-in access controls. “Access controls” decide who, or which agent, can see each piece of data. AWS Context is built on the same base as Amazon’s Quick assistant.
AWS Continuum: Catching Security Holes in Code
The second service is called AWS Continuum. It deals with security. To be exact, it handles “vulnerabilities” in code. A vulnerability is a weak spot. An attacker could use it to break in.
AWS Continuum tries to do the whole job from start to finish. First, it finds the weak spots. Then it ranks them by how much they could hurt the business. Next, it checks if each one can really be used by an attacker. It does this test in a separate, walled-off space. So the real system stays safe. Last, it suggests an exact fix.
AWS Continuum uses special security models to do this. For now, it is not open to everyone. Only a few chosen test customers can use it. This early test group is called a “pilot.”
Key Facts: The Two New AWS Services
| Service | What it does | Status |
|---|---|---|
| AWS Context | Builds a knowledge graph from company data so agents know real business facts and make fewer mistakes | Generally available (built on the same base as Amazon’s Quick assistant) |
| AWS Continuum | Finds, ranks, tests, and fixes security weak spots in code automatically | Available only to select pilot customers |
Other Updates Announced at the Summit
AWS shared a few more updates with the two main services. The AWS DevOps Agent got new features. They are called Release Readiness Review and production-like testing. These check code changes before they go live. This is in preview and free in the US East region.
AWS also launched Kiro as a native iOS app. Now people can review and approve code from a phone. And Bedrock AgentCore grew too. It got a managed knowledge base. It also got new connectors to S3, SharePoint, Confluence, and Google Drive. Plus, it added security filters from partners like Check Point and Zscaler.
Why It Matters (Especially for India and Founders)
Many Indian startups and developers build on AWS. So these tools could reach you soon. If you run AI agents, two things have held you back. One is wrong answers. The other is security worries. These two services aim right at both.
For founders, business context is the bigger story. An agent that knows your real data is far more useful than one that guesses. That can mean better support bots and faster internal tools. For developers, automatic security checks save time. The hard part is finding and fixing weak spots. That is exactly what AWS Continuum targets.
The bigger picture is worth watching too. AI costs money to run. It needs memory and chips. You can see this in Anthropic and Micron’s memory deal. And the choice of AI model keeps changing. Options like GLM-5.2 are in the mix. AWS is making a bet. It thinks context and safety will decide who wins, not just raw model power.
FAQ
What are the two new AWS AI agent services?
They are AWS Context and AWS Continuum. AWS Context gives agents business knowledge. AWS Continuum finds and fixes security weak spots in code.
Why does AWS say AI agents need these?
AWS says agents often miss real business context. So they guess and make mistakes. Agents also act fast. They move faster than old security tools can keep up with.
Can anyone use them right now?
AWS Context is generally available. That means anyone can use it. AWS Continuum is open only to select pilot customers for now. That means a small, early test group.
What is a knowledge graph in simple words?
It is a map of facts that shows how things connect. For example, it links a customer to their orders. The agent reads it to know the truth. So it does not have to guess.
The Takeaway
AWS is making a clear point. To make AI agents truly useful, they need to know your business. And they need to stay safe. AWS Context handles the knowledge. AWS Continuum handles the security. Together, they target the two weak spots that hold agents back today.
