AWS re:Invent 2025 — Day 1: The Pace Picks Up

Day 1 of AWS re:Invent 2025 — announcements, conversations, community, and the unexpected moments that make this event what it is.

Day 1 at re:Invent is when the the event finally shifts from anticipation to full throttle. The Community Lounge wakes up early, the first wave of announcements rolls in, hallway conversations get louder and more speculative, and the Expo floor turns into a kind of organised chaos that only AWS could orchestrate. Today felt like the moment everyone collectively remembered why we fly across the world for this: the people, the pace, the surprises, and the unmistakable sense that the week is about to go somewhere big.

Image

I started my morning the same way I always do here: breakfast in the Community Lounge, notebook open, coffee in hand, and a quiet half hour to catch up on whatever the previous night delivered. Overnight there had been another burst of announcements, and while people were still buzzing about Lambda on EC2, the consensus remains somewhere between confusion and curiosity. It clearly fills an enterprise-scale pricing and control gap, but without GPU support (yet) it is struggling to win hearts. Maybe this week will reveal the real plan behind it.

The networking has already become one of my favourite parts of the conference. There is something both amusing and completely normal about flying across the world only to run into the same Australians I last spoke to in this exact building, one year ago. Conversations ranged from tech deep dives to community initiatives to the realities of local markets where some regions are heating up and others are fighting to keep momentum. And of course, the re:Invent staple: speculation. What is coming in the keynotes? Which rumours will prove true? Which will quietly disappear?

With that, let’s get into the announcements.

Announcements Through the Night

AWS AI League 2026 Championship

AWS has introduced a new global competition with a 50,000 USD prize pool, giving builders a chance to compete using services like Amazon Bedrock AgentCore and Amazon SageMaker AI. The format feels spiritually aligned with DeepRacer: equal parts learning, engagement and community bragging rights. I can already see AWS User Groups turning this into friendly rivalry for the next 12 months.

URL: https://aws.amazon.com/about-aws/whats-new/2025/11/ai-league-2026-championship/

AWS Interconnect (Preview)

AWS has taken a significant step into a world most enterprises already live in: multicloud. AWS Interconnect is a managed private connectivity service that lets you create high speed network connections between your AWS VPCs and your VPCs in other public clouds, as well as back to your own offices and data centres. No more stitching things together with third-party ISPs and bespoke routing if you do not want to.

It comes in two flavours:

  • Interconnect Multicloud
    Link your AWS VPC to your environments in other cloud providers. This is a major step forward for organisations running distributed architectures, active active DR strategies or simply choosing the right cloud for the right workload. The reality is most enterprises are already multicloud, whether by design or accident. Interconnect finally meets them where they are, with private connectivity owned by AWS rather than a patchwork of carriers.

  • Interconnect Last Mile
    Connect your AWS VPC back to your on-premises environment or office network. At first glance it feels like Direct Connect made easier. I want to dig into this, because if it simplifies some of the complexity and friction around Direct Connect, that will be a welcome shift for a lot of customers.

URLs:
https://aws.amazon.com/about-aws/whats-new/2025/11/preview-aws-interconnect-multicloud/
https://aws.amazon.com/about-aws/whats-new/2025/11/gated-preview-interconnect-last-mile/

Multimodal Retrieval for Amazon Bedrock Knowledge Bases (GA)

Amazon Bedrock Knowledge Bases now support retrieval across text, images, audio and video. Until now, customers were limited to text documents and images, which often ruled Knowledge Bases out of more complex enterprise RAG workflows. This update means you can ask questions like “show me Q1 projections for Amazon Bedrock” and have the system pull relevant material from documents, charts, recorded meetings, training videos and audio clips in one go.

I have seen several use cases dismissed because customers did not want to build bespoke ingestion or retrieval systems just to handle non textual formats. This change makes Knowledge Bases far easier to use in the real world for customers who want a managed service rather than a build it yourself solution.

URL: https://aws.amazon.com/about-aws/whats-new/2025/11/multimodal-retrieval-bedrock-knowledge-bases/

Amazon CloudWatch Incident Reports with Five Whys

Amazon CloudWatch incident reports can now guide teams through a Five Whys analysis directly inside the service. It is modelled on Amazon’s own correction of errors process and is meant to help teams arrive at a more structured root cause.

For incident response and managed services teams this could be a useful addition, especially for those who do not already have a mature post incident review process. Like any AI assisted workflow though, it is still worth reading the output carefully before accepting it as gospel.

URL: https://aws.amazon.com/about-aws/whats-new/2025/11/amazon-cloudwatch-incident-reports-five-whys-analysis/

Announcements Overnight Worth a Mention

Amazon SageMaker Catalog Automatic Data Classification

Amazon SageMaker Catalog now provides automated data classification using AI agents. It suggests business glossary terms as data is published, reducing the manual effort of tagging datasets and improving metadata consistency across large organisations. It is not a headline grabber, but it is the kind of quality of life improvement that starts to matter at scale.

URL: https://aws.amazon.com/about-aws/whats-new/2025/11/amazon-sagemaker-catalog-automatic-data-classification-ai-agents/

Amazon EKS Capabilities

Amazon EKS has announced the general availability of EKS Capabilities, a fully managed, extensible set of Kubernetes native platform features for workload deployment, cloud resource management and resource composition. In practice it means more of the platform glue that usually lives in operators, controllers and pipelines can now be offloaded to AWS.

This is squarely aimed at teams who are all in on Kubernetes and want more of the platform heavy lifting taken care of. I will leave the excitement to the people running clusters all day; someone is very happy about this.

URL: https://aws.amazon.com/about-aws/whats-new/2025/11/amazon-eks-capabilities/

A Wave of Amazon Connect Updates

Amazon Connect continues to evolve as the customer contact platform of choice. In one burst it received:

  • AI powered case summaries
  • Support for multiple knowledge bases, including integration with Amazon Bedrock Knowledge Bases
  • Simplified linking of related contacts and cases using flows
  • AI powered predictive insights (preview)
  • Custom visual themes in the agent workspace
  • Streaming of messages for AI powered interactions
  • Agent initiated chat workflows
  • Automated performance evaluations for self service interactions

It is remarkable how far Connect has come from its early days as “a contact centre in the cloud”. It is now a deeply integrated, data driven platform that is getting a very solid glow up to close out 2025.

AWS Transform Goes Big

On Monday, AWS Transform quietly went from preview curiosity to a fully fledged, generally available product family. For many enterprise customers this might be one of the most important sets of announcements this year.

Transform is essentially AWS’s agentic AI powered approach to crushing technical debt across Windows, mainframe, VMware and large legacy codebases. It is aimed at the repeatable, cross codebase migration tasks that usually soak up 20 to 30 percent of engineering effort and are nobody’s favourite work.

There are a few key strands:

  • AWS Transform Custom: Organisation specific code and application modernisation at scale, using agentic AI to automate version upgrades, runtime migrations, framework transitions and even language translations.
  • Full stack Windows modernisation: Modernise .NET Framework and .NET code to newer targets such as .NET 10 or .NET Standard, including porting ASP.NET Web Forms to Blazor and migrating Entity Framework code. The new developer experience inside Visual Studio is more interactive, with editable transformation plans, estimated effort, real time progress and handoff notes for AI code companions.
  • Transform for mainframe: New reimagine capabilities and automated test planning and generation to support mainframe migration projects. Test plan generation, test data collection and automation scripts are all aimed at accelerating the testing side of modernisation, which is often where mainframe projects stall.

Transform has not always had the warmest reception. Last year’s preview landed with more raised eyebrows than adoption. But this GA release feels far more complete and far more confident. AWS is making a clear statement here: they believe this is a real path forward for customers drowning in legacy, and they are ready to invest in making it work at scale.

For organisations struggling with shrinking mainframe skill sets, legacy Windows estates or just years of accumulated technical debt, this could become a very important tool in the kit.

The Expo Opens

When the Expo Hall opened, the crowd control improvements were immediately noticeable. AWS has clearly learned from previous years and adjusted the flow to avoid the Day 1 bottleneck. The doors opened, a cheer went up and this time I was standing on the inside at the Thoughtworks booth.

I spent the afternoon from 4 to 6 pm at the booth supporting our demo for the year: “Ideas to Impact with AI”. It is an AI racer challenge where you navigate across a city, connecting certain geographical landmarks while trying to be more efficient than the AI route. Think of it as optimising your own delivery network and seeing whether you can outsmart the model.

It landed well. We had a steady queue and roughly one person every three minutes stepping up to play. It does not hurt that there is a prize draw for a pair of Meta AI glasses for participants. Incentives help.

Thoughtworks: AWS Partner of the Year (Data and Analytics)

For me personally, the standout announcement of the day did not come from a service page. Thoughtworks was named AWS Partner of the Year in the Data and Analytics category.

This is a huge achievement that reflects years of work across teams, regions and customer engagements. It is one of those moments where the long running effort around skills, delivery and community really shows up in a single line of recognition.

We held a networking reception to celebrate. There were a lot of congratulations, a lot of good questions and, as usual, far more people asking about the golden jacket than I expected. Conversations that start with “how many certifications is that” tend to end with “how do we build stronger community around this”, which is exactly how I like it.

A great Day 1. The announcements are rolling, the Expo is humming, the community is in full voice and the week feels like it is just getting started.