If you were paying attention to the “What’s Next with AWS 2026” event last week, you already know something shifted. If you didn’t follow it, I’ll catch you up — because this matters more than the usual release cycle.
AWS announced three things that, taken individually, sound like product updates. Taken together, they look like a strategic repositioning of the entire AWS ecosystem around AI.
Amazon Quick now has real surface area
Amazon Quick — which you might still remember as QuickSight, AWS’s BI tool — has been quietly evolving into something much broader: an AI work assistant available across all your devices. Last week it landed with a desktop app for macOS and Windows (Preview), a free plan that doesn’t require an AWS account, and a new feature called Dataset Q&A that deserves attention.
The desktop app matters because it means Quick can now access your local files, calendar, and communications without needing a browser tab. It’s a real shift: from “BI dashboard in the browser” to “ambient AI assistant on your machine”.
Dataset Q&A is the feature that interests me most for development audiences. It’s powered by a text-to-SQL agent that interprets natural language questions, identifies the correct data source, and generates precise SQL in a single conversational step — working against Redshift, Athena, Aurora PostgreSQL, and Apache Iceberg tables on S3. No need to build topics or dashboards first. You ask, it queries, you get an answer.
S3 table buckets are now a supported data source, meaning you can run conversational analytics directly on your data lake — without intermediate data warehouse or OLAP layers.
On integrations: Quick expands its native connections to include Google Workspace, Zoom, Airtable, Dropbox, and Microsoft Teams. A direct move into territory occupied by tools like Notion AI, Glean, and Microsoft 365 Copilot.
The OpenAI partnership is the headline enterprise teams will take seriously
AWS and OpenAI announced a deeper partnership that places OpenAI’s frontier models directly on Amazon Bedrock infrastructure. It’s in limited preview, but it’s worth understanding the mechanics now.
GPT-4o and GPT-4 are going to be available through the same Bedrock APIs that teams already use — with unified security, governance, and cost controls. No separate infrastructure to set up, no new security model to learn.
But the detail that will change budget conversations: Codex in Bedrock lets you authenticate with AWS credentials, process inference through Bedrock infrastructure, and apply Codex usage toward existing cloud commitments with AWS.
If your company has an Enterprise Discount Program or committed spend agreement with AWS, this means OpenAI’s code agent potentially doesn’t come out of a separate AI tools budget — it comes out of the cloud spend you’re already committed to. That’s a concrete internal procurement argument for teams that had been holding back on Codex adoption.
Codex in Bedrock is available through the Bedrock API and works with the Codex CLI, Codex desktop app, and VS Code extension.
How this looks from a CTO/CIO perspective
AWS isn’t competing with individual AI tools anymore. It’s building an answer to the question every enterprise tech leader is asking right now: “How do I consolidate my AI spend without losing model quality?”
The answer AWS proposes: run everything through Bedrock. You get model diversity (Anthropic, Meta, Mistral, now OpenAI), unified governance, and everything counts against your existing cloud commitment. Amazon Quick gives your non-technical teams an AI assistant connected to the same data infrastructure. Amazon Connect agents handle customer-facing workflows.
It’s a bet that companies will prioritize simplification over selecting the best tool in each category. Given how much AI tool sprawl grew over the last 18 months, it’s not an unreasonable bet.
What it means for Latin American teams
Some points worth noting in our context:
Amazon Quick’s free plan is genuinely accessible — you sign up with a Google, Apple, GitHub, or Amazon account, no AWS account needed. It significantly lowers the barrier for teams and individuals wanting to explore the tool without going through enterprise procurement.
Access to Codex in Bedrock and GPT-4o is in limited preview with no announced timeline for regional availability. Latency from South American locations to US-East AWS infrastructure is real — something to consider before building workflows that depend on these APIs.
The data sources supported by Dataset Q&A (Redshift, Athena, Aurora PostgreSQL, S3 Tables) are well-suited to teams already running native AWS data infrastructure. If your stack lives elsewhere, this feature in particular won’t move the needle for you.
The bottom line: if your organization already has cloud commitments with AWS, this week’s announcements give you concrete arguments to consolidate more of your AI spending within that existing relationship. If you’re cloud-agnostic, the story is less immediately compelling — but the direction AWS is moving is clear, and it’s worth watching closely.