AWS Re:Invent 2023 Recap: Big Announcements, Gen AI & Frugal Architects

Post • 6 min read

The curtains have closed on another AWS re:Invent in Las Vegas. And so the cloud community - us included - end 2023 with excitement over the announcements and insights gathered across the event.

Maarten, Antti and Mariusz - a few of our AWS guys that attended the event - have shared their pick of the key takeaways, announcements and the best sessions to catch up on if you missed them.

The big theme: Generative AI (and natural language) everywhere

Among the announcements, the main theme was the presence of Generative AI across the board, with the integration of natural language into diverse AWS services.

It was also visible everywhere, not only during keynotes, but also through the expo - on probably 75% of stands

Amazon Q

The most exciting announcement was probably Amazon Q. AWS CEO Adam Selipsky announced the release, which is a collection of features and services combined under the same umbrella.

On one side Q can, for example, support builders and developers with things like troubleshooting network connectivity, finding the right EC2 instance for your particular workload or transforming application code transformation when moving to newer versions.

On the other hand Q can support employees across the company by providing answers on business questions that require information spread across different business applications like Slack and Confluence.

Amazon Bedrock enhancements

A few enhancements were announced for Amazon Bedrock. One of them was Guardrails - allowing users to set the limits on what’s used by AI models. So if queries are irrelevant to a bot, you can prevent responses, instead of letting it generate inaccurate responses. You can also define topics that bots aren’t not allowed to touch and set filters for specific phrases and words.

Sagemaker boost

There are new capabilities for SageMaker to help users improve building, training, and deployment of foundation models. The HyperPod is expected to cut times for training foundational models by up to 40%, enabling distributed training at scale. And SageMaker Canvas can help improve data preparation through natural language instructions.

FMOps & LLMOps - A new need

A lot of customers are rightly excited about these large language models (LLMs) and about Gen AI as a game changer for their businesses. But turning these cool tools into everyday business operations isn't like flicking a switch. You need to understand how to make Gen AI applications a part of regular operations - which leads to something now being called FMOps (foundation model operations).

And many are zooming in on one of the most common uses of Gen AI: text, and the operations around large language models - LLMOps. Businesses need to think about the processes involved, the people using them, how to pick and evaluate these models, and dealing with data privacy. It's really about getting these advanced AI models to work out there in the real world.

A route to market through marketplace

Cloud marketplaces are expected to see incredible growth again next year (we’ve written an article about this here) - and AWS partners have already introduced more than 90 gen AI demos and 50 partner solutions in the AWS Marketplace.

And AWS announced some changes to the AWS Marketplace, including the SaaS Quicklaunch. This is like a shortcut that lets partners easily set up and start using AWS with ready-made templates covering things like the security standards.

AWS is also making things more convenient with AWS Marketplace APIs for sellers, so buyers don't have to leave their website to make a purchase. And AWS is lowering its prices in the AWS Marketplace, so depending on the deal size, could save buyers between 1.5% to 3%.

The Frugal Architect

Werner Vogels, Amazon's CTO, emphasised the need for businesses to be cost-conscious in managing cloud expenses during his keynote [watch here] at AWS re:Invent. And, with cloud spending projected to increase significantly, Vogels urged techies in particular to rediscover the art of architecting for cost - what he calls being the Frugal Architect.

Aligning cost with business goals, optimising incrementally, and monitoring operations were key recommendations. AWS also unveiled tools like myApplications and CloudWatch Application Signals to help enterprises manage costs effectively. And Vogels also spoke about how being cost-conscious not only saves money but also contributes to sustainability.

“Security is job zero”

Ensuring the safety of data and systems is the absolute top priority in cloud, and AWS reiterated this point. It's like saying before we do anything else, we must make sure everything is secure and protected. So, security isn’t just a task among many; it's the very first and most important thing to take care of. It’s built into everything to keep customers and partners safe.

Other interesting announcements

Other interesting announcements, generally were improvements to existing services. Some of our favourites are:

AWS Console-to-Code

[Learn more about it here] 

Over the years there have been great open-source projects that would achieve the same result, converting something you “clicked” together in the AWS Management Console into AWS CloudFormation templates. Now AWS is integrating this functionality directly, with EC2 as the service to support it. Once this matures it will be a time-saver to support the conversion of a quick prototype build in the console to infrastructure-as-code.

Amazon RDS for DB2

[Learn more about it here]

Looking at the ranking of database engines, DB2 was the next major DB engine that was not available as managed service on AWS yet.

AWS CloudFormation introduces Git management of stacks

[Learn more about it here]

GitOps for AWS CloudFormation, this feature allows you automatically deploy and update CloudFormation stacks from a connected Git repository.

Amazon CloudWatch announces AI-powered natural language query generation (in preview)

[Learn more about it here]

This feature enables you to easily and quickly generate queries in context of your logs and metrics data using plain language. For example “show me all failed login attempts”.

Various improvements to Amazon CodeWhisper

[Learn more about them here]

Amazon CodeWhisperer now supports AI-powered code generation for CloudFormation and Terraform. It can now also identify and remediate security issues in your code. Combing it with Amazon Q you can also build new applications from descriptive prompts or let Q explain what a certain section of code does.

Things that surprised us

Amazon One Enterprise

[Learn more about it here]

A new Amazon service that uses hand palms for biometric identification both for physical access such as secure areas in buildings, but also as a replacement for multi-factor authentication in Enterprise applications.

Amazon WorkSpace Thin Clients

[Learn more about it here]

Amazon’s stepping into end-user computing hardware, namely a thin client that can be used to access Amazon Workspaces and AppStream.

Not as many announcements as you’d expect

There was actually a relatively low number of announcements. Especially during the keynotes from Peter DeSantis, where there were perhaps only five. So it feels like an  evolution, not revolution - which is in contrast to other providers.

Not all about the tech

The focus moved from technology to how AWS makes life easier or better for people and customers. And this doesn’t mean better in a technical manner for things like usability, management, cost control - and more green.

Our favourite sessions

ARC307 Do modern cloud applications lock you in?

[Watch it here]

Gregor Hophe is a well-known Enterprise Strategist, now turned Evangelist. In this talk he guides the audience through what aspects and considerations come into play when determining the lock in when choosing a particular cloud platform. Quick teaser; there is not only vendor lock-in.

The Frugal Architect, Dr Werner Vogels

[Watch it here]

Mentioned above, Werner talked about “the frugal architect” and how you should architect systems with cost in mind.

AIM245 Innovate faster with generative AI

[Watch it here]

Dr. Bratin Saha, VP of AWS AI and ML Services at AWS, talks about what it takes for Enterprise to build an scale using generative AI.

DOP309-R AI for DevOps: Modernising your DevOps operations with AWS

[More info here]

AWS showed how tools like Amazon CodeWhisperer, Amazon CodeGuru, and Amazon DevOps Guru use smart technology (like AI) to automatically spot and fix problems and keep things running smoothly when using microservices.

HYB207-INT Innovation talk: Emerging tech

[Watch it here]

AWS highlighted the unique ways companies are using cloud: from quantum computing and IoT to genomics, drug discovery, airplanes, and space stations.

Get in Touch.

Let’s discuss how we can help with your cloud journey. Our experts are standing by to talk about your migration, modernisation, development and skills challenges.

Maarten Bruntink LinkedIn
CTO, AWS
Antti Hannula LinkedIn
AWS Global Practice Lead
Mariusz Preiss
Mariusz Preiss LinkedIn
Product Lead - Landing Zones