#711: AWS News: Claude 3.7 Meets Amazon Bedrock, Bedrock Data Automation

Summary

AWS Podcast Episode #711 Summary: Claude 3.7 Meets Amazon Bedrock, Bedrock Data Automation

Release Date: March 10, 2025
Hosts: Alicia and Gillian Ford
Podcast: AWS Podcast

---

Introduction and Upcoming Mailbag Episode

In Episode 711 of the AWS Podcast, hosts Alicia and Gillian Ford kick off the show with enthusiasm, even managing a smooth broadcast despite operating in a two availability zone mode due to Shruti's absence. Alicia announces an upcoming Mailbag episode, inviting listeners to submit their burning questions about AWS services, the cloud, GenAI, and more through awspodcastamazon.com. She emphasizes the community-driven nature of the podcast, encouraging listeners to share questions that might benefit others as well.

Alicia [00:00]: "We have something in the works... we want you to send us your burning questions about AWS services, the cloud, Genai, whatever's on your mind to awspodcastamazon.com."

Claude 3.7 and Amazon Bedrock Integration

The episode delves into one of the standout topics: Anthropic's Claude 3.7 and its integration with Amazon Bedrock. Gillian exclaims the significance of this development, highlighting its impact on AWS customers.

Gillian Ford [01:15]: "It is totally a big deal. I'm really excited for customers to use this."

Alicia elaborates on the innovations Claude 3.7 brings, particularly its hybrid reasoning model that combines quick responses with complex problem-solving within the same model call. This integration simplifies processes by eliminating the need for separate calls for different types of answers.

Alicia [01:23]: "This represents a change in the way that a Frontier model operates in that it's a hybrid reasoning model... it can do the quick response and the complex problem solving in the same model in the same call."

Gillian adds her appreciation for the standard and extended thinking modes, which allow customers to balance latency and model self-reflection based on their specific use cases.

Gillian Ford [01:53]: "Being able to test those so you can be able to better understand what's the right approach is going to be really awesome for customers."

Bedrock Data Automation

Shifting focus to Bedrock Data Automation, the hosts discuss its transition to general availability, emphasizing its role in streamlining data extraction from unstructured multimodal documents such as PDFs, images, audio, and video files.

Gillian Ford [03:41]: "This one streamlines the process of extracting insights from unstructured multimodal documents... it can be a PNG, your JPEGs, your audio files, your video files."

Alicia highlights the feature’s flexibility, allowing businesses to create blueprints tailored to their specific schemas or formats, facilitating seamless integration with downstream systems.

Alicia [04:06]: "You can also create these things called blueprints, which are sort of tailored to your business needs... which again helps with that integration piece."

She shares a personal anecdote about the challenges of ensuring consistent JSON outputs from language models, underscoring how Bedrock Data Automation’s blueprints can mitigate such issues.

Alicia [04:06]: "...with Bedrock Data Automation, you're actually just doing this in just like one API call. So this could be one way of being able to reduce your data extraction costs totally."

Comprehensive AWS Product Updates

The majority of the episode is dedicated to a broad overview of the latest AWS service updates, categorized for clarity:

1. Analytics

• Amazon Managed Streaming for Apache Kafka (MSK): Now supports Apache Kafka version 3.8, introducing features like compression level configuration for optimized performance.
• Fine-Grained Access Control: Enhanced via AWS Lake Formation with EMR on EKS, simplifying access management.
• Dynamic EMR Cluster Updates: Allows real-time reconfiguration without terminating or restarting clusters, minimizing downtime.

2. Artificial Intelligence

• Amazon Bedrock Session Management APIs: Now in preview, these APIs simplify state and context management for GenAI applications using frameworks like Langgraph and Llama Index. They enable secure management of multi-step workflows, checkpointing, and enhanced debugging.

Gillian Ford [03:57]: "This is really useful for securely isolating those particular sessions and supporting multi tenant applications."

• Amazon Q Developer in SageMaker Canvas: Now generally available, offering enhanced ML use cases, improved data analytics, and faster response times.
• Bedrock Guardrails Expansion: Increased service quotas, allowing up to 50 calls per second on the Apply Guardrail API and an eightfold increase in content filtering capabilities.
• Microsoft 365 Integration: Amazon Q Business now integrates with Word and Outlook, expanding third-party data source support.

3. Compute (EC2 and ECS)

• Amazon EKS Anywhere: Extended support for Kubernetes versions 1.28 and above.
• Amazon EC2 Fleet Enhancements: Added support for block device mapping overrides, enabling broader EC2 capacity access and parameter customization per instance type.
• Time-Based Copy for AMIs: General availability of time-based copy for Amazon Machine Images, allowing customers to specify desired completion durations (15 minutes to 48 hours) for AMI copy operations.
• AWS Batch Resource Aware Scheduling: Introduces resource-aware token management to reduce job failures and optimize compute resource utilization.
• Amazon Elastic Beanstalk Updates: Now supports Corretto 17 and 21 with Tomcat 11, Ruby 3.4 on Amazon Linux 2023, and increased CPU limits for ECS tasks up to 192 VCPUs.

Alicia [11:49]: "That's a lot of VCPUs."

Gillian Ford [11:51]: "Totally. I want to know what people are going to do with it."

4. Customer Engagement

• Amazon Connect Enhancements:
  • Shift Exchange for Agents: Allows agents to swap shifts, enhancing schedule flexibility.
  • Telephony Price Reductions in Vietnam: Significant price cuts on direct inward dial and outbound minutes.
  • Interactive Welcome Messages: Now supported for initiating chats.

Alicia [13:57]: "There's your hot tip of the day. Change the certificates before they run out to save a lot of pain."

5. Databases

• Amazon Timestream for InfluxDB: Added read replica support to scale read operations across multiple instances and availability zones.
• Database Insights for RDS MySQL and MariaDB: Offers on-demand performance analysis, identifying discrepancies from normal performance and providing corrective advice.
• Amazon RDS Enhancements:
  • IAM DB Authentication Metrics and Logs Visibility.
  • Support for New Minor Versions: Including spatial patch bundles for Oracle.
  • Managed Blue/Green Deployments: Simplifies database upgrades with minimal downtime.

Alicia [15:58]: "It's usually either really badly written SQL or an index that you dropped or forgot to recreate."

6. Developer Tools

• AWS CodeBuild Updates:
  • Support for macOS 15.2 builds.
  • Managed webhooks for GitHub Enterprise.
  • Managed runners for GitLab self-managed instances.
  • Enhanced compute options and parallel test report merging.
• AWS CodePipeline Enhancements:
  • New console experience for viewing pipeline releases.
  • Native deployment support for Amazon EC2 and Amazon EKS.

7. End User Computing

• Amazon AppStream 2.0: Improved user experience with support for certificate-based authentication on multi-session fleets running Windows OS joined to an Active Directory.

8. Front End Web and Mobile

• Amplify Hosting Updates:
  • Support for IAM roles in server-side rendered (SSR) applications.
  • Secure credential management with AWS Secrets Manager and Parameter Store.
  • Granular permission controls and branch-specific role configuration.
• Amazon Location Service: Now supports AWS PrivateLink, enhancing security by keeping API access within the VPC.

9. Internet of Things (IoT)

• AWS IoT Device Defender Enhancements: Improved certificate lifecycle management with new audit checks for certificate age and expiration, preventing service disruptions due to expired certificates.

10. Management and Governance

• AWS CDK Updates:
  • L2 construct support for Amazon Kinesis Data Firehose delivery streams, enabling infrastructure as code for streaming data.
• AWS Price List API: Now supports AWS PrivateLink.
• Backup Payment Methods for Invoices: Allows alternate payment methods to ensure timely invoice payments without manual intervention.
• Application Signals: Enables centralized monitoring across multiple AWS accounts with a single-pane view.

11. Migration and Modernization

• AWS Database Migration Service (DMS):
  • Supports network interfaces and secure connection types.
  • Assigns private IP addresses for migrations, enhancing security.
• DMS Serverless: Now includes pre-migration assessments to identify potential migration issues in advance, reducing delays.

12. Security, Identity, and Compliance

• AWS WAF Updates:
  • Enhanced integration with service quotas.
  • Expanded data protection with new controls for sensitive data in logs.
  • Improved logging configuration for optimal logging options.
• AWS Network Firewall Enhancements:
  • Simplified policy management with enhanced console features.
  • Introduction of automated domain lists and insights based on HTTP/HTTPS traffic logs.
• Amazon Verified Permissions: Now supports the Cedar JSON entity format, aligning with the open-source Cedar SDK for streamlined authorization requests.

13. Storage

• Amazon Data Lifecycle Manager: Now supports AWS PrivateLink, ensuring API access within the VPC and enhancing security for storage operations.

Highlighted Listener Insights

As the hosts navigate through the extensive updates, they emphasize the practical implications for AWS users. Alicia humorously relates the importance of regular database patching to everyday tasks, reinforcing best practices in cloud management.

Alicia [16:02]: "It's like being told to clean your room. Sometimes you stand back and look at your room go yeah, I do clean your desk is probably what I do."

Gillian underscores the significance of Claude 3.7, labeling it as the most anticipated feature among listeners.

Gillian Ford [22:25]: "I mean, the Sonnet 3.7 I would say is like the big one. That's what people have been really waiting for."

Conclusion and Call to Action

In their closing remarks, Alicia expresses excitement about integrating Bedrock Session Management APIs into her projects, highlighting the relief of automating complex context management tasks.

Alicia [22:32]: "It sounds like we've got some undifferentiated heavy lifting taken care of."

Gillian provides contact information for listeners to reach out, encouraging ongoing engagement.

Gillian Ford [22:59]: "Jillian Ford on LinkedIn."

Alicia reiterates the invitation for listeners to submit questions for future episodes and signs off with an encouragement to keep building.

Alicia [23:01]: "And of course, until next time, keep on building."

---

Key Takeaways:

• Claude 3.7's integration with Amazon Bedrock offers a unified model handling both quick responses and complex problem-solving, enhancing developer efficiency.
• Bedrock Data Automation simplifies data extraction from diverse document types, providing customizable blueprints for seamless integration with existing systems.
• AWS continues to expand and enhance services across analytics, AI, compute, databases, developer tools, and more, focusing on scalability, security, and user-friendly features.
• Listeners are encouraged to engage with the podcast by submitting questions and interacting with hosts on LinkedIn.

For more detailed updates and to stay informed about the latest AWS advancements, subscribe to the AWS Podcast and join the community conversation.

Transcript

[Alicia]

This is episode 711 of the AWS podcast, released on March 10, 2025. G'day everyone and welcome back to the AWS Podcast. I'm Alicia with it. Great to have you back, joined by one of my illustrious co hosts, Gillian Ford. G'day, Gillian. How you going?

[Gillian Ford]

I'm always excited to be here. Simon, what about you?

[Alicia]

I am excited to be here too. We're operating in two availability zone mode at the moment because Shruti's not with us, but because we have three AZs, the show always goes on, which is good. And referring to the show, we've got something in the works. In fact, we love hearing from you and hearing what's on your mind, et cetera. So we want you to send us your burning questions about AWS services, the cloud, Genai, whatever's on your mind to awspodcastmazon.com and we're thinking about doing a kind of a mailbag episode upcoming. So this is a chance to get some insights, some questions answered that may have been on your mind. So just like I said, just drop us a note AWS podcast. Com and the chances are if you've got a question, lots of other people do too. So it's always good to share that. But there's been some lots of interesting things going on, but we've picked out a couple from the bunch. Gillian anthropic Claude 3.7 kind of a big deal.

[Gillian Ford]

It is totally a big deal. I'm really excited for customers to use this. Simon, what excites you about 3.7?

[Alicia]

Well, I think this represents a change in the way that a Frontier model operates in that it's a hybrid reasoning model and what that means is basically it can do the quick response and the complex problem solving in the same model in the same call. So previously you would sort of call things separately to get different types of answers. This kind of wraps it all into one good thing, which makes it easier. What about yourself? I mean there's a bunch of changes. What's slipped out at you?

[Gillian Ford]

Yeah, I really like the standard and extended thinking modes because a lot of customers are really thinking about latency versus really giving the model the extra time for self reflection and what's the best for their use case. And so I think being able to test those so you can be able to better understand what's the right approach is going to be really awesome for customers.

[Alicia]

Agree. I also like that there's more token output links, so it's up to 128k in beta 64k generally available. Because certainly what I'm finding is that, you know, when you need to do lots more work or do longer code reviews and stuff like that, suddenly start to run out of output. So having that get bigger is important. I think the other really interesting thing is that we had this model available one day after it was released in Bedrock. So again, knowing and using and loving Bedrock means that you get access to all the cool models when they come out.

[Gillian Ford]

It's just crazy how fast I mean, that would take for just like the team of people we've got working on this behind the scenes.

[Alicia]

Yeah, it takes work. And like, you know, Deepseek 1 came out a little while ago, that was the sort of the big hotness in the industry and everyone was talking about it. And what was cool is that almost immediately we had obviously the instructions to be able to import it yourself, because that's another thing that Bedrock can do.

[Gillian Ford]

Absolutely, yeah. I love that customers are really interested in different options, especially those that really want to test all of the latest in open source. So this one's really exciting for those.

[Alicia]

Yeah, I think it's. I think it's a cool one. Now, speaking of generative AI, as we seem to do a lot these days, you can't have it without data. Data is kind of important. And Amazon Bedrock Data Automation went generally available. So remind us what that was, because this popped up last year in preview, didn't it?

[Gillian Ford]

Yeah. So this one streamlines the process of extracting insights from unstructured multimodal documents. So think your like PDF files, this could be an image or PNG, your JPEGs, your audio files, your video files. So this is really going to just simplify that entire process of being able to extract data from these types of data sets.

[Alicia]

Yeah, And I think what it does is it. It helps you obviously do things out of the box. So it understands documents and videos and stuff like that. But you can also create these things called blueprints, which are sort of tailored to your business needs. So if you know a particular schema or format that you need to get it into for downstream systems, so sort of maybe JSON, maybe xml, goodness knows what, you can pre program it to do that for you, which again helps with that integration piece. That's one of the things that I know I've been struggling with sometimes with some of the models is depending on how you prompt them and how you manage them, you're not always going to get exactly the same answer. Like, I had one recently where it's like I told it, hey, provide me with a JSON output for such and such, and it was doing a tremendous job and then suddenly decided that on certain inference it would say, here's your JSON string, and then it would give me the JSON. Now, of course, my code hadn't anticipated the commentary before the JSON string. And so I was blowing up going, oh, there's no, you're not giving me JSON here. It's like, well, I told the model not to. I was like, do not give me anything other than JSON. So I think all of these blueprints will help too.

[Gillian Ford]

Yeah, I think so. I'm sure a lot of people can relate to that experience as well. Or even those who maybe already are doing some data extraction using just an LLM. So they might be thinking like, all right, so why would I take what I'm already doing and just now use Bedrock Data Automation? So one of the reasons might be, let's say you're having multiple prompts to be able to extract the data. Like there might be different things that you're having to ask the large language model. Well, with Bedrock Data Automation, you're actually just doing this in just like one API call. So this could be one way of being able to reduce your data extraction costs totally.

[Alicia]

That's a really good call. And you're processing time as well, so you sort of. It's a double win. Well, let's talk about some of the other things that have been going on out there in AWS land. Let's start with the topic of analytics and Amazon managed streaming for Apache Kafka. Amazon MSK now supports Apache Kafka version 3.8, so you have lots more new options. Key new features include support for compression level configuration, which means you can optimize performance when using different compression types, and lots of other cool things. We're also announcing fine grained access control via AWS lake formation with EMR on eks. No, I can't fit any more acronyms into that, but it's kind of important because it basically means that you can simplify your access control and only give access to things that people need. You can now also dynamically update your running EMR cluster with reconfiguration, for instance fleets, so you don't have to terminate the cluster or do a restart. You can just dynamically update it, which is cool because who's got time for outage time? Now let's talk about artificial intelligence. Amazon Bedrock has Launched Session Management APIs for Genai Applications in preview. Now this is a new capability that enables developers to simplify state and context management for Genai applications built with popular open source frameworks like Langgraph and Llama index session management APIs provide out of the box solution that enables developers to securely manage state and conversation context across multi step gen AI workflows. Which means you don't have to do it yourself, which I'm kind of happy about because I always end up coding that myself. By preserving this session state between interactions, it means things like virtual assistants and multi agent research. Workflows that need the persistent context between extended interactions are just there and you can use this capability to checkpoint workflow stages. You can save intermediate stages, resume tasks from point of failure or interruption. You can also pause and replay sessions and leverage detail traces to debug and enhance your own application. And by treating session as a first class resource it means you can also have granular access control through IAM and you can encrypt the data using KMS as well. So this is really useful for securely isolating those particular sessions and supporting multi tenant application. So I'm pretty excited about this. I'm going to have to have to work that into something I'm doing at the moment. Amazon Q Developer is now generally available in Amazon SageMaker Canvas so you now get that access to that development capabilities including support for additional ML use cases, enhanced data analytics and faster response times. And Amazon Bedrock Guardrails now has an increase in service quota limits so it is now a 2x increase. You can now process up to 50 calls per second on the Apply Guardrail API and there is an 8 times increase on the Content filters, sensitive filters and Word filters as well. And AWS has announced Microsoft 365 for Word and Outlook integration into Amazon Q Business. So these let you use Q Business across those particular data sets. So this continues to grow the number of third party data sources that you can plug queue for business into. And AWS Chatbot is now named Amazon Q Developer so this sort of cleans up the naming a little bit for you. With this launch Microsoft Teams and Slack Chat application names will change to Amazon Q Developer and the notifications you'll get will now have the Amazon Q mention instead of the W s, but it still does all the cool stuff.

[Gillian Ford]

Now let's talk about computer AWS is announcing extended support for Kubernetes versions for Amazon EKS anywhere and this is for versions 1.28 and above. Amazon EC2 fleet has added support for block device mapping overrides so customers use EC2 flight to access wider EC2 capacity across instance types and Availability zones within a single launch request. So now you can override the values specified for select parameters in the launch template for each instance type, and with this release customers can also override block device mapping parameters specified in the launch template. Amazon EC2 announces the general availability of time based copy for Amazon Machine Images, similar to time based copy for EBS Snapshots. This feature enables customers to meet their compliance objectives by ensuring that AMIs are copied within and across AWS regions within a specified duration. So previously customers could not predict or control the duration of their AMI copy operations. But now with this capability, customers can specify a desired completion duration. This could range from 15 minutes to 48 hours for individual AMI copy requests. AWS Batch now supports Resource Aware scheduling so with Resource Aware scheduling you can set up sets of tokens representing these resources which will then be consumed by the running AWS batch jobs. This will help reduce job failures and wasted compute time caused by missing or rate limited resources, which in turn will improve utilization of infrastructure and reduce costs. Amazon Elastic Beamstock now supports Windows Server 2025 and Windows Server Core 2025 environments. AWS Elastic Beanstalk now supports Corretto 17 and 21 with Tomcat 11. AWS Elastic Beanstalk now Supports Ruby 3.4 on Amazon Linux 2023. Amazon ECS now supports CPU limits of up to 192 VCPUs for ECS tasks deployed on EC2 instances. This is an increase from the previous release.

[Alicia]

That's a lot of VCPUs.

[Gillian Ford]

Totally. I want to know what people are going to do with it. Hopefully you'll email us an increase from the previous of 10 VCPU limit.

[Alicia]

That's a big increase as well. It's not just a big number. It's like someone's asked for this.

[Gillian Ford]

Yes, this enhancement allows customers to more effectively manage resource allocation on larger Amazon EC2 instances. Amazon ECS adds support for additional IAM condition keys. These new condition keys let you create IAM policies as well as service control policies to better enforce your organizational policies in containerized environments. With this release, ECS has added condition keys that allow you enforce policies such as ECS Task cpu, ECS Task Memory, and Tag Propagation are just some examples. Next up we've got Customer engagement. Amazon Connect now allows agents to exchange shifts with each other, providing a greater schedule flexibility without compromising service levels. Amazon Connect reduces telephony pricing in Vietnam. This includes price reductions of direct inward dial minutes by 95% from 0.0815 per minute to 0.004 per minute. Wow, that's a pretty big production. And outbound minutes by 44% from 0.0896 per minute to 0.05 per minute. Amazon Connect now supports interactive welcome messages when starting chats. Amazon SES announces that outbound customers can now specify a mail Manager archive resource as an additional destination for outbound mail workloads. This enables retention of messages by post dkim signature, ensuring that the archive is usable for validating every individual sent message. The MailManager archive search interface allows easy discovery of index messages and presents search results directly in the AWS console.

[Alicia]

Let's talk about databases Amazon timestream for InfluxDB has added read Replica support, so now you can scale your read operations across multiple instances and and availability zones and you can do it right from within the console as well. So this is really useful to support high read throughput so you have more instances to cope with the read traffic while still having a single write endpoint Database Insights now provides on demand analysis for RDS MySQL and RDS MariaDB, so you can now analyze database performance for a time period of your choice. You can learn how the selected time periods differ from normal, what went wrong, and get advice on corrective actions. That's pretty cool. And you can also understand what's contributing to your performance issues. It really makes diving deep into performance issues a lot easier. Amazon RDS now provides visibility into IAM DB authentication metrics and logs, and Amazon RDS for MariaDB now supports new minor versions, your regular reminder to update your stuff and Amazon RDS for Oracle now supports Spatial patch bundle for January 2025 Spatial Patch and CloudWatch data database insights has now added support for RDS databases as well, so again, another way to quickly get access to troubleshooting information, dive deep and fix stuff. Spoiler alert. It's usually either really badly written SQL or an index that you dropped or forgot to recreate. Just putting that out there. Some more updates in terms of versions. Amazon RDS for PostgreSQL has new minor version support and Amazon RDS for MySQL has other new versions as well. If using RDS, it is easy to upgrade and I encourage you to do so. You can also test your upgrades, you can also test your application against the new version of the database, et cetera. But it's really important to stay current both from a security perspective and also to get the added goodness and performance improvements that you get from these updates. And you can use the Amazon RDS Managed Blue Green deployments for simple upgrades as well.

[Gillian Ford]

Simon, it doesn't get old, you telling everyone to go patch their stuff.

[Alicia]

Well, I'll stop saying it when people do it, we all need to remind it's like being told to clean your room. Sometimes you stand back and look at your room go yeah, I do clean your desk is probably what I do.

[Gillian Ford]

That's true. Now let's talk about developer tools. AWS Code build now supports macOS 15.2 as a new major version for macOS builds AWS CodeBuild adds support for managed webhooks in GitHub Enterprise. AWS CodeBuild adds support for managed runners for GitLab self managed AWS CodeBuild now supports merging parallel test reports and new compute options. This enhancement also allows you to select a mix of on demand instances, reserve capacity fleets or lambda compute resources for your parallel tests. AWS CodePipeline introduces a new console experience for viewing pipeline releases. AWS CodePipeline adds native Amazon EC2 deployment support and last AWS CodePipeline adds native Amazon EKS deployment support. Next up is end user computing. Amazon AppStream 2.0 improves the end user experience by adding support for certificate based authentication on multi session fleets running the Microsoft Windows operating system and joined to an active directory.

[Alicia]

Let's talk about front end web and mobile Amplify Hosting has announced support for IAM roles for server side rendered or SSR applications. So this allows you to enable secure connections to other address resources from server side rendered applications. So this unlocks a few new things. So firstly, now have secure credential management for SSR applications, direct integration into AWS services like Secrets Manager and parameter store support for database connections like RDS and DynamoDB, granulous permission controls and branch specific role configuration. So this is a big deal if you build those kinds of apps and the Amazon Location service now supports AWS privately. Quick update on the topic of the Internet of Things. We're happy to announce new features for AWS IoT device Defender to improve certificate lifecycle management. Always fun to do a new audit check for certificate age and enhancements to the existing device. Certificate expiring Audit check check makes life easier to see when things are running out before they run out. There's your hot tip of the day. Change the certificates before they run out to save a lot of pain. Let's talk about management and governance. AWS CDK releases L2 construct support for Amazon data firehose delivery streams. So this means you can define and deploy your streaming data infrastructure as code and you can programmatically configure your streams that automatically deliver real time data to destinations like Amazon S3. The AWS Price List API now supports AWS PrivateLink and we're also happy to announce backup payment methods for invoices. So this feature lets you have an alternate payment method that will be automatically charged if your primary payment method fails. So this is really useful for making timely invoice payments without the need for manual intervention or last minute payment updates. You can now monitor and observe apps across multiple accounts with application signals, so this gives you a centralized single pane of glass tracking for all your systems across the board. And you can now use your China UnionPay credit card to create an AWS account, so this eliminates the need for international credit cards for customers in China.

[Gillian Ford]

Two Updates in Migration and Modernization AWS Database Migration service now supports network interfaces. Customers can choose a secure connection type which improves connectivity to on premises databases for DMS homogeneous migrations for each data migration, DMS now assigns a private IP address. Additionally, a non static public IP address may be assigned if the connection is configured for public access. This access can be established within the local VPC or through VPC peering, Direct Connect or vpn. AWS Database Migration Service Serverless now supports pre migration assessments for replications. A pre migration assessment evaluates the service and target database of a database migration task to help identify problems that might prevent a migration from running as expected, this seems super useful. By identifying and fixing these issues before a migration starts, you can avoid delays in completing the database migration.

[Alicia]

Now let's talk about security, identity and compliance. A couple of updates for AWS WAF firstly, it enhances integration with service quotas, which means you can proactively monitor and manage your quotas for your cloud deployments. And it has also expanded its data protection capabilities with new controls for sensitive data in logs. And we've also updated the logging configuration console experience, making it easy for you to select the optimum logging option. The AWS Network Firewall has simplified policy management with again enhanced console features so easier ways to do things. And it also now introduces automated domain lists and insights. So the new capability analyzes HTTP and HTTPs traffic logs from the last 30 days and provides insights into frequently accessed domains, which means you can create rules quickly based on observed network traffic patterns. And Amazon Verified Permissions now supports the Cedar JSON entity format. Developers can use this simpler format for authorization requests and this aligns Amazon Verified Permissions API more closely with the open source Cedar SDK and makes it easy to move from the SDK to Amazon verified permissions or vice versa. And one quick update for the topic of storage. Amazon Data Lifecycle Manager now supports AWS Private Link A few services this week getting into that particular support capability. Again, having Private Link means that your API access is all happening in your VPC instead of connecting over the Internet. So that's it. Lots of interesting updates. Julian. Anything that really leaps out to you or something, you think, oh, I've got to give that a go.

[Gillian Ford]

I mean, the Sonnet 3.7 I would say is like the big one. That's what people have been really waiting for.

[Alicia]

That's the one. That's the one.

[Gillian Ford]

What comes to mind for you?

[Alicia]

I'm probably going to look at the session management thing because I've been doing some very hacky stuff in terms of passing context to different agents. I'm big in the multi agent world now. I've moved on from I don't make one call to my LLM. I make lots of calls to my LLMs now. So, you know, it's all about things working together and orchestrating that can be tricky. But it sounds like we've got some undifferentiated heavy lifting taken care of. So Gillian, how do folks reach out to you?

[Gillian Ford]

Jillian Ford on LinkedIn.

[Alicia]

And if you want to go old school, awspodcastmazon.com is the place to do it. Remember, we also want your questions for the Mailbag episode as well. So if there's something in the back of your mind you've always wanted to task, go right ahead. And of course, until next time, keep on building.