Amazon Web Services (AWS) unveiled a set of new tools designed to accelerate the development and deployment of agentic AI systems, marking a significant expansion of its AI infrastructure portfolio. The announcements center on enabling enterprises and developers to build, scale, and secure intelligent agents across industries.
At the core of the launch is Kiro, a purpose-built integrated development environment (IDE) for agentic AI. Kiro introduces innovations such as spec-driven development and intelligent agent hooks that streamline how developers write and test code for autonomous AI systems. Complementing this, Bedrock AgentCore provides enterprises with a framework to securely operate highly capable AI agents at scale, integrated into AWS’s existing Bedrock ecosystem for model deployment.
AWS also introduced Strands Agents, an open-source toolkit that supports popular frameworks such as A2A and MCP, giving developers flexibility to build and extend agent workflows. The AWS Marketplace now includes listings of AI agents and tools from providers such as Anthropic, Salesforce, PwC, Accenture, and IBM, expanding customer options for agent deployment. Together, these services position AWS to compete at the forefront of the agentic AI wave.
• Kiro: agentic IDE with spec-driven development and agent hooks
• Bedrock AgentCore: enterprise-grade framework for secure and scalable AI agents
• Strands Agents: open-source toolkit supporting A2A and MCP frameworks
• AWS Marketplace expanded with AI agents from Anthropic, Salesforce, PwC, IBM, and others
• Positioned to support enterprises building autonomous AI workflows at scale
“Our conviction that AI will change every customer experience is starting to play out… launched AI models like DeepFleet that optimize productivity for our 1M+ robots, made it much easier for developers to write code with Kiro, launched Strands to make it easier to build AI agents, and released Bedrock AgentCore to enable agents to be operated securely and scalably,” said Andy Jassy, President and CEO of Amazon.
🌐 Why it Matters: The release of Kiro, Bedrock AgentCore, and Strands signals AWS’s push to define the developer stack for agentic AI, a category likely to dominate enterprise AI adoption. By combining secure infrastructure with open frameworks, AWS is positioning itself to capture workloads as enterprises move beyond
🌐 What is: A2A (Agent‑to‑Agent Protocol):
Agent‑to‑Agent (A2A) emerged in early 2025 as an open protocol developed by Google and formally hosted by the Linux Foundation to enable secure, interoperable communication between autonomous AI agents . The project launched with backing from over 50 major enterprise and tech partners—including Atlassian, Box, Cohere, Salesforce, SAP, Intuit, Workday, ServiceNow, PwC, Deloitte, and Accenture—which collectively aim to standardize how agents discover, negotiate and delegate tasks across heterogeneous systems . A2A provides a peer‑to‑peer communication framework built on common protocols like HTTP, JSON‑RPC, and server‑sent events to ensure agents can interoperate across vendor ecosystems without exposing their internal state or requiring custom bridges .
🌐 What is: MCP (Model Context Protocol):
The Model Context Protocol (MCP) originated in November 2024 when Anthropic—developers of the Claude models—released it as an open‑source spec to solve the “N×M” integration problem between AI models and external tools and data sources . Anthropic’s vision positioned MCP as a “USB‑C for AI,” creating a standard, JSON‑RPC‑based interface for LLMs to access context, invoke functions, or read data from systems like Slack, GitHub, or Postgres without bespoke connectors . Soon after launch, major platforms such as OpenAI and Google DeepMind adopted MCP, and early MCP servers appeared across tools like Zed, Sourcegraph, Replit, Block, and Apollo, supported by a growing ecosystem of SDKs and reference implementations .