AWS Builder Center
AWS Summit New York 2026: The Agentic AI Era Has Arrived (Part 1 of 2)

AWS Summit New York 2026: The Agentic AI Era Has Arrived (Part 1 of 2)

Part 1 of the covering announcement at AWS Summit

On June 17, 2026, AWS packed the Jacob K. Javits Convention Center in Manhattan for its flagship East Coast developer conference. The keynote, led by Dr. Swami Sivasubramanian (VP of Agentic AI) and Chet Kapoor (VP of Security Services and Observability), made one thing unmistakably clear: AWS is betting its platform future on autonomous AI agents that build, secure, operate, and optimize software with minimal human intervention.
This is Part 1 of a two-part series covering every major announcement. Here we focus on the developer tooling and agent infrastructure side of things. Part 2 covers security, data, business intelligence, and operational agents.


1. Kiro Pro Max and Kiro Mobile for iOS

What it is: Kiro, AWS's agentic IDE built on spec-driven development, received two major upgrades at the summit. Kiro Pro Max is a new tier with enhanced capabilities, and Kiro Mobile is a native iOS app that lets developers manage agent-driven coding sessions from their phone.
Kiro works differently from traditional AI coding assistants. Instead of reacting to individual chat prompts, it works from structured specifications — a requirements document and a design document are produced before code generation begins. The agent then plans, implements, and verifies changes across the entire codebase.
The iOS app connects to cloud-based agent sessions. You can start work, monitor progress, review diffs, and approve changes remotely without needing your laptop running.
Use Case: A lead architect is commuting on the subway when her team's Kiro agent finishes implementing a multi-service refactor she specified that morning. She opens Kiro Mobile, reviews the diff against the spec, approves two of three task groups, and requests changes on the third — all before reaching her office. The agent incorporates feedback and has a passing build ready by the time she sits down. No laptop required for the approval workflow.

2. Amazon Bedrock AgentCore Harness (GA)


What it is: The managed agent harness in Amazon Bedrock AgentCore is now generally available. AWS describes it this way: if the foundation model is the brain, the harness is the body — everything the brain needs to get work done. It provides identity management, memory, tool access through the gateway (with security policies enforced), organizational knowledge integration, web search, and observability. Every agent action is governed and traced from the first call without additional wiring.
Teams go from idea to working agent in minutes rather than weeks of infrastructure assembly.
Use Case: A fintech company wants to build a compliance-checking agent that reads new regulatory documents, cross-references them with existing internal policies, and flags gaps. With AgentCore Harness, the team defines the agent's tools (document retrieval, policy database queries, notification APIs), and the harness handles authentication, session memory across multi-turn conversations, audit logging, and scaling — all managed. The team ships a production-ready compliance agent in days instead of building custom orchestration infrastructure.

3. Web Search on Amazon Bedrock AgentCore

What it is: A fully managed tool that enables agents to ground their responses in current, cited web knowledge. It uses a built-in connector on the AgentCore Gateway via Model Context Protocol (MCP). The agent sends a natural-language query, and Web Search returns relevant snippets, source URLs, titles, and publication dates. Crucially, there is zero data egress from the customer's secured AWS environment.
Use Case: An enterprise customer support agent needs to answer questions about competitor product changes, industry news, or recently published API documentation. Rather than relying solely on a static knowledge base that becomes stale, the agent calls Web Search to retrieve the latest information, cites its sources in the response, and the customer gets accurate, up-to-date answers. The security team is satisfied because all data stays within the AWS environment boundary.

4. Amazon Bedrock Managed Knowledge Base


What it is: A new managed primitive that abstracts away the multiple infrastructure components developers traditionally assemble for enterprise RAG (Retrieval-Augmented Generation) pipelines — storage, retrieval, embeddings, re-ranking, and foundation model selection — into a single service. It includes native data connectors, Smart Parsing for automatic multi-format data preparation, and an Agentic Retriever for complex multi-step queries. It integrates directly with AgentCore Gateway.
Use Case: A healthcare organization wants to build a clinical decision support agent that queries across thousands of medical journals, internal research notes (PDFs, DOCX, HTML), and structured drug interaction databases. Previously, this required stitching together S3, OpenSearch, embedding models, a re-ranker, and custom chunking logic. With Managed Knowledge Base, the team points connectors at their data sources, Smart Parsing handles the diverse formats automatically, and the Agentic Retriever decomposes complex clinical questions into sub-queries. Time-to-production drops from months to weeks.

5. AgentCore Optimization Capabilities

What it is: New optimization capabilities in AgentCore that turn production traces into continuous improvement for agents. AWS highlighted that the most dangerous agent failures are not the ones that throw errors — they are the silent ones that look fine on dashboards but produce subtly wrong outputs. The optimization system analyzes real production execution traces to identify degradation patterns and opportunities for improvement.
Use Case: An e-commerce company's product recommendation agent starts producing lower conversion rates, but no errors appear in monitoring. The AgentCore optimization system detects that a tool the agent relies on began returning slightly different response formats after a partner API update. It surfaces the drift, suggests a prompt adjustment, and the team restores recommendation quality without debugging from scratch.

6. AWS Transform — Continuous Modernization (Preview)


What it is: An autonomous agent service that continuously scans code repositories to detect, prioritize, and remediate technical debt at scale. Out of the box, it includes policies for detecting end-of-life dependencies, deprecated frameworks, and other common sources of technical debt. It generates findings in hours and can autonomously submit pull requests with fixes.
Industry analysts called this "perhaps the most interesting idea from AWS Summit 2026" — a service designed to continuously modernize legacy infrastructure including mainframes and .NET applications.
Use Case: A large bank runs 200+ microservices, many of which were built 3-5 years ago on frameworks that are now approaching end-of-life. Their platform team enables AWS Transform, which scans all repositories overnight. By morning, it has identified 847 findings across 140 services, prioritized by risk (security vulnerabilities from deprecated dependencies rank highest), and already submitted 23 pull requests upgrading the most critical libraries. The platform team reviews and merges rather than manually hunting for outdated packages across hundreds of repos.

What Ties It All Together

These six announcements form a coherent vertical stack for the agentic development lifecycle:
  1. Kiro handles how agents (and human-AI teams) specify and build software
  1. AgentCore (Harness + Gateway + Knowledge Base + Optimization) handles how agents run, remember, authenticate, retrieve knowledge, and improve in production
  1. AWS Transform handles continuous maintenance of the code those agents produce
The message from the keynote is that AI agents are no longer experimental. They are production systems that need the same rigor — specifications, observability, security, continuous improvement — that we expect from any critical infrastructure.
Continue to  Part 2  for coverage of AWS Continuum, AWS Context, AWS DevOps Agent, Amazon Quick, and AWS WAF AI Traffic Monetization.
Sources:
Give likes and subscribe
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article