AWS taking advantage automatic logicWhich uses math-based validation to build new capabilities in its Amazon Bedrock AgentCore As a platform, the company is diving deeper into the agentic AI ecosystem.
Announced during its annual ReInvent conference in Las Vegas, AWS is adding three new capabilities to AgentCore: "Policy," "Evaluation" And "episodic memory." The new features aim to give enterprises more control over agent behavior and performance.
AWS also revealed what it called “a new class of agents.”" Or "frontier agent," Which are autonomous, scalable and independent.
Swami Sivasubramanian, AWS’s vice president for agentic AI, told VentureBeat that many of AWS’s new features represent a shift in who the builder becomes.
“We’re really on the cusp of a major tectonic shift with AI, but agentic AI is really starting to change the art of what’s possible, and that’s going to make it one of the most transformative technologies,” Sivasubramanian said.
policy agent
new PPolicy capability helps enterprises reinforce guidelines even if the agent has already given its feedback.
David Richardson, AWS’s vice president for AgentCore, told VentureBeat that the policy tool sits between the agent and the tools it calls, rather than baked into the agent, as fine-tuning often is. The idea is to prevent an agent from violating enterprise rules and redirect it to reevaluate its reasoning.
Richardson gave the example of a customer service agent: A company will write a policy saying the agent can give a refund up to $100, but for more than that, the agent must return the customer to a human being. He said it remains easy to break an agent’s reasoning loop through, for example, quick injection or poisoned data, causing the agent to bypass guardrails.
“There are always quick injection attacks where people try to distort the agent’s logic to make the agent do things it shouldn’t do,” Richardson said. “That’s why we implemented the policy outside of the agent, and it works using the automated reasoning capabilities we’ve spent years helping customers define their capabilities for.”
AWS unveiled automatic logic checking At Bedrock at last year’s Re:Invent. These use neurosymbolic AI or mathematics-based verification to prove correctness. This tool applies mathematical proof to models to confirm that it is not a hallucination. AWS is leaning heavily into neurosymbolic AI and automated reasoning, emphasizing enterprise-grade safety and security in ways that differ from other AI model providers.
Episodic memories and evaluation
Two other new updates to AgentCore, "Evaluation" And "episodic memory," Also gives enterprises a better view of agent performance and gives agents episodic memory.
AIn the enhancement of agentcore memory, episodic memory refers to knowledge that agents use only occasionally, as opposed to long-term preferences, which they have to constantly refer back to. Context window limitations hinder some agents, so they sometimes forget information or conversations they haven’t used for a while.
“The idea is to help get the information the user would actually want the agent to remember when they come back," Richardson said. "For example, ‘What is their favorite seat on the plane for family trips?’ or ‘What kind of price range are they looking for?’"
Episodic memory differs from the previously mentioned AgentCore memory in that instead of relying on maintaining short- and long-term memory, agents built on AgentCore can recall certain information based on triggers. This may eliminate the need for custom instructions.
With AgentCore Assessment, organizations can use 13 pre-built assessors or create their own. Developers can set up alerts to warn agents if they start failing quality monitoring.
frontier agent
But perhaps AWS’s strongest push into enterprise agentic AI is the release of Frontier Agents, or fully automated and independent agents, who the company says can act as teammates with less direction.
The concept is similar, if not identical, to competitors’ more asynchronous agents. Google And OpenAIHowever, it appears that AWS is releasing more than just autonomous coding agents,
Sivasubramaniam called him "new class" of agents, "Not just a step-by-step action change in what you can do today; They range from assisting with individual tasks to complex projects."
The first is Kiro, an autonomous coding agent that has been in public preview since July. At the time, Kiro was introduced as an alternative to Vibe coding platforms like OpenAI manual Or windsurfSimilar to Codex and Google’s numerous asynchronous coding agents, including julesKiro can independently code, review, fix bugs and schedule the tasks he needs to complete.
Meanwhile, AWS Security Agent embeds deep security expertise into applications from the start. The company said in a press release that users “define security standards once and the AWS Security Agent automatically validates them across your applications during review – helping teams address the risks that are critical to their business, not a generic checklist.”
AWS DevOps agents will proactively help developers, especially callers, find system breaks or bugs. It can respond to events using its knowledge about the application or service. It also acknowledges the relationship between the application and the tools it uses, such as Amazon CloudWatch, Datadog, and Splunk, to find the root cause of the problem.
Enterprises are interested in deploying agents and, ultimately, bringing more autonomous agents into their workflows. And, while companies like AWS continue to provide security and control to these agents, organizations are slowly figuring out how to connect it all.
<a href