We sent a team to the Amazon Web Services (AWS) Summit which took place in New York City on June 17. The summit is one of their largest annual events, where cloud and AI practitioners gather to see what’s next. The agenda was packed with over 200 sessions covering everything from agentic AI to data strategy to code modernization.
We came back with pages of notes, a lot of excitement, and a few things we think are worth sharing — whether you’re a tech veteran or someone who’s just starting to wonder what all this AI talk actually means for your business.
Here’s our plain-English breakdown of the biggest takeaways.
AI Is Moving from Chatbots to Agents — and That’s a Big Deal
A lot of people are familiar with AI chatbots: you ask a question, you get an answer. What dominated the conversation at this summit was something a step further — AI agents.
An agent doesn’t just answer questions. It takes actions. It can look up information in your databases, draft a document, query your CRM, pull a report, and hand off results to another agent — all on its own, with minimal human hand-holding.
Amazon is betting heavily on this shift with a platform called Amazon Bedrock AgentCore, which became generally available in late 2025. Think of it as the infrastructure layer that lets you build, deploy, and manage these agents securely — at any scale. The important thing for non-technical readers: you’re not locked into Amazon’s AI models. AgentCore works with Claude (from Anthropic), OpenAI, Google Gemini, Meta’s Llama, and others. You pick what works best for your needs.
Alongside AgentCore, AWS released an open-source toolkit called Strands SDK — essentially the coding toolkit developers use to actually write the agent logic before deploying it on AgentCore. The two work together like a blueprint and a construction site.
Your Data Is the Foundation — and Most Organizations Aren’t Ready
This was perhaps the most honest, ground-level conversation of the day. Speaker after speaker made the same point: the quality of your AI is only as good as the quality of your data.
That sounds obvious, but the implications run deep. Before you can build a useful AI agent, you need to answer some hard questions:
- Is your data clean, or is it full of duplicates, outdated records, and gaps?
- Is it organized in a way a machine can understand?
- Do you have good metadata — descriptions, tags, categories — so the AI knows what it’s looking at?
The recommended pipeline is straightforward but requires real investment: Raw data → Clean data → Enriched data → Ready for AI. Skipping steps leads to what one presenter bluntly called “data failure” — an agent that confidently gives you wrong answers because it’s working from bad inputs.
For publishing and content-heavy businesses, this is especially relevant. Unstructured content (PDFs, Word docs, files in Dropbox or SharePoint) needs to be properly chunked, tagged, and indexed before an AI can search it meaningfully. The term for this process is RAG (Retrieval-Augmented Generation) — and getting it right is what separates a useful AI assistant from an unreliable one.
Amazon Quick: The AI Workspace You Probably Haven’t Heard Of
One of the most interesting products shown was Amazon Quick — Amazon’s all-in-one AI-powered workspace for employees. Imagine a smart assistant that can search across all your company’s tools and data simultaneously, in plain English.
It connects to 40+ business systems out of the box: Salesforce, Slack, SharePoint, Gmail, Dropbox, Smartsheet, ServiceNow, and many more. Ask it a question and it pulls answers from across all of them — respecting your existing security permissions so people only see what they’re supposed to.
One feature worth calling out specifically: Amazon Quick in QuickSight lets you build business intelligence dashboards and reports just by describing what you want in plain English. No SQL. No data analyst is required for routine reports. That’s a meaningful shift for smaller teams.
AWS Transform: Getting Your Legacy Code Up to Speed
A less flashy but very important announcement: AWS Transform has been expanded significantly. This is Amazon’s AI-powered tool for modernizing old codebases — automatically.
If your organization is running on aging software, outdated programming languages, or legacy infrastructure, AWS Transform can analyze your code and handle a lot of the migration work that would normally take months of expensive developer time. By their numbers: over 4.5 billion lines of code analyzed since launch, saving customers more than 1.6 million hours of manual effort.
The new capability announced at the NYC Summit is continuous modernization — an AI agent that monitors your codebase on an ongoing basis and flags (or fixes) technical debt as it accumulates, rather than waiting for a big migration project every few years.
For development teams under pressure to ship new features while keeping existing systems stable, this is the kind of tool that can quietly save a lot of headaches.
What This Means for Westchester Publishing Services
We didn’t go to this summit just to learn — we went with our own operations in mind. Here’s where we see the most relevance for what we do:
Knowledge management: Tools like Amazon Quick could make it far easier to search and surface information across our own documents, files, and internal systems — saving time that currently goes into manual navigation and retrieval.
Workflows: AI agents that can pull data, generate status updates, and route tasks between systems.
Code and system health: For any development work we maintain, AWS Transform’s continuous modernization capabilities are worth a close look.
Data readiness: Perhaps most importantly, the summit reinforced that the organizations getting the most from AI are the ones who invest in clean, well-structured, well-tagged data. That’s the homework that pays off later.
Final Thoughts
The theme running through nearly every session we attended was this: AI is only as smart as the information you give it, and only as useful as the problems you point it at. While the technology is advancing rapidly, many organizations are still working toward the strategic clarity needed to determine where to start.
As we continue to explore these tools, we continue to be guided by our core principles:
- Protecting the security of each title and client’s IP
- Always maintaining high quality and accurate output
- Ensuring transparency, choice, and scalability for our clients
- Always keeping a human in the loop
Keep an eye on this space and our LinkedIn page as we share more information about technological developments and how they can become part of publication workflows, following the core principles outlined above. If you’d like to discuss how Westchester can help you with any aspect of your editorial or production workflow, contact us to have a conversation with a member of our team.
