Google and AWS split the AI agent stack between control and execution

Source: VentureBeat

Overview

The era of enterprises stitching together prompt chains and shadow agents is nearing its end as more options for orchestrating complex multi‑agent systems emerge. As organizations move AI agents into production, the question remains: how will we manage them?

Google and Amazon Web Services offer fundamentally different answers, illustrating a split in the AI stack. Google’s approach runs agentic management on the system layer, while AWS’s harness method sets up management in the execution layer.

The debate on how to manage and control gained new energy this past month as competing companies released or updated their agent‑builder platforms—Anthropic with the new Claude Managed Agents and OpenAI with enhancements to the Agents SDK—giving developer teams options for managing agents.

• AWS added new capabilities to Bedrock AgentCore, optimizing for velocity by relying on harnesses to bring agents to product faster, while still offering identity and tool management.
• Google’s Gemini Enterprise adopts a governance‑focused approach using a Kubernetes‑style control plane.

Each method offers a glimpse into how agents move from short‑burst task helpers to longer‑running entities within a workflow.

Upgrades and Umbrellas

Google Gemini Enterprise

Google released a new version of Gemini Enterprise, bringing its enterprise AI agent offerings—Gemini Enterprise Platform and Gemini Enterprise Application—under one umbrella.

• Vertex AI has been rebranded as Gemini Enterprise Platform. Aside from the name change and new features, the underlying interface remains the same.
• Maryam Gholami, senior director of product management for Gemini Enterprise, explained:

“We want to provide a platform and a front door for companies to have access to all the AI systems and tools that Google provides. The way you can think about it is that the Gemini Enterprise Application is built on top of the Gemini Enterprise Agent Platform, and the security and governance tools are all provided for free as part of the Gemini Enterprise Application subscription.”

AWS Bedrock AgentCore

AWS added a new managed agent harness to Bedrock AgentCore. According to a press release shared with VentureBeat, the harness “replaces upfront build with a config‑based starting point powered by Strands Agents, AWS’s open‑source agent framework.”

• Users define what the agent does, the model it uses, and the tools it calls.
• AgentCore stitches everything together and runs the agent.

Agents Are Now Becoming Systems

The shift toward stateful, long‑running autonomous agents has forced a rethink of how AI systems behave. As agents move from short‑lived tasks to long‑running workflows, a new class of failure emerges: state drift.

• Agents accumulate state—memory, responses, evolving context.
• Over time, that state can become outdated as data sources change or tools return conflicting responses.
• The agent becomes more vulnerable to inconsistencies and less truthful.

Reliability therefore becomes a systems problem. Managing drift may require more than faster execution; it may need visibility and control. Platforms like Gemini Enterprise and AgentCore aim to prevent this failure point.

Gholami noted that customers will dictate how they want to run and control any long‑running agent:

“We are going to learn a lot from customers where they would be using long‑running agents, where they just assign a task to these autonomous agents to just go ahead and do. Of course, there are tricks and balances to get right and the agent may come back and ask for more input.”

The New AI Stack

It’s becoming clear that the AI stack is separating into distinct layers that solve different problems.

• AWS, and to a certain extent Anthropic and OpenAI, optimize for faster deployment. Claude Managed Agents abstracts much of the backend work for standing up an agent, while the Agents SDK now includes support for sandboxes and a ready‑made harness. These approaches lower the barrier to getting agents up and running.
• Google offers a centralized control panel to manage identity, enforce policies, and monitor long‑running behaviors.

Enterprises likely need both approaches. Practitioners must have a serious conversation about how much risk they are willing to take.

“The main takeaway for enterprise technology leaders considering these technologies at the moment may be formulated this way: while the agent harness vs. runtime question is often perceived as build vs. buy, this is primarily a matter of risk management. If you can afford to run your agents through a third‑party runtime because they do not affect your revenue streams, that is okay. On the contrary, in the context of more critical processes, the latter option will be the only one to consider from a business perspective,” — Rafael Sarim Oezdemir, head of growth at EZContacts (VentureBeat).

Iterating quickly lets teams experiment and discover what agents can do, while centralized control adds a layer of trust. Enterprises need to ensure they are not locked into systems designed purely for a single way of executing agents.