← AWS Podcast
AWS Podcast
#742: Amazon Quick Suite, AWS MCP, and lots more!
00:30:35
Stay up-to-date, with almost 60 new updates this week!!
Loading summary
Transcript62 lines
- [00:00]
A
This is episode 742 of the AWS podcast released on October 20, 2025.
- [00:09]
B
Hello everyone and welcome back to the AWS Podcast. Simon here with you. Great to have you back, joined by my co host, Gillian Ford. G', day Gillian. How you going?
- [00:16]
A
G', day, Simon. It's always a good day when it's podcast day.
- [00:19]
B
This is true. This is true. Lots of new updates, just a touch under 60 today. That's right, six zero. And there's a couple that really bear talking about early on. And we'll talk about some during the episode as well. But the is that we're introducing Amazon QuickSuite, which is your agentic AI powered workspace. So this is in general availability. It's a new set of agentic teammates that help you get the answers you need, using all of your business data to move instantly from insights to action. So QuickSuite retrieves insights across the public Internet and your documents, including information in popular third party applications, databases and other places. Your company keeps important data. So if you need a single data point or a research project, I did one recently, I'll talk about that shortly, or anything in between. This helps you get started up and running. It also can do things as well. So it can create or update JIRA tickets or servicenow incidents. It can help you automate a lot of routine tasks. So things like responding to RFPs or preparing for meetings, et cetera. The nice thing is all your data is safe and private, so your queries and data are never used to train models. Really important. And you can tailor the QuickSuite experience to your environment as well. There's also a 30 day free trial for up to 25 users as well. We've been using this ourselves, Gillian, as it's sort of been been coming to fruition. I've used the research side of things and I found it really quite useful because the research process, it's. It's kind of familiar to folks. If you use something like perplexity, et cetera, it's sort of useful in terms of gathering lots of broad data around the place. What's nice is it produces a report, but then in the side you can put comments. It's like, it's almost like, you know, when someone sends you a document, you do your edits and your comments in the document, but you do that in the document and then you say, now go do the research again. And it will action those points, which is really cool. Like I found it kind of, it felt humanistic in doing it that way because like I'd done a piece of domain research in a particular area and it took a particular direction and I wanted to take a different direction. And so, like, in just a few points, I said, you know, this should be more about, you know, the Australian experience, or this should be more about this, et cetera. Funny that. And it, and it responded in that next iteration going, oh, yeah, I'll update it doing that way. So that's kind of what I've used it. Have you used it for much, Gillian, or still. Still exploring.
- [02:45]
A
I'm definitely still exploring it for the research purposes of taking an Excel spreadsheet and being able to just really understand trends, especially across, at least for me, with healthcare and AI, and really trying to be able to understand that. So I'm definitely excited for that and especially on the automation part of being able to automate routine tasks.
- [03:06]
B
Yep, that's. That's definitely something I need to look more deeply in because again, it's, it's interesting as sort of Cambrian explosion of. Of AI stuff is, there's lots of different ways to do things and, and what's sort of the way this week may not be the way next week. So kind of true.
- [03:21]
A
That's a good point.
- [03:23]
B
Now, another new announcement. Funnily enough, AI related is a really handy one because I've got a tip that I hope is going to make people really happy. So it's the AWS API MCP Server version one release. So this is our model Context protocol, MCP Server, enabling foundation models to interact with any AWS API through natural language by creating and executing syntactically correct CLI commands. That phrase again. Syntactically correct CLI commands. The reason why I'm excited about this is. I don't know about you, but a lot of the LLMs that I use are pretty good at using AWS, but more often than not, they issue the CLI command they think they should issue.
- [04:07]
A
Yes.
- [04:07]
B
Then they get an error and it's.
- [04:10]
A
Like, wait, you can't error?
- [04:11]
B
Yeah, it's like you made the error and then they fix it. So it's like, oh, you know, they'll often they'll use a flag, like a true or false flag. Drives me insane. And I actually reached out to some colleagues who do a lot of work in this space and said, this is driving me nuts. What do I do? And they're like, oh, you should just change it to tell it. So instead of, for example, within. Within qcli, for example, I tell it. So instead of using the Use AWS tool to Use the Call AWS tool now. The Call AWS tool is part of this AWS API MCP and it gets it right every time. It's awesome. And so it really helps a lot in terms of getting things working. It also has access to something called a Suggest AWS command tool as well. This allows it to figure out what it should be doing. Also has improved security, file system controls, better input validation. You can also use the AWS CloudWatch agent to get logs from it as well. For observability. It also now offers streamable HTTP transport in addition to stat IO and it also includes elicitation in supported MCP clients as well to improve that human in the loop workflow. You can also deny certain actions as well. You can also get There's a new experimental tool called Get Execution Plan that will tell you what it would do for common tasks as well. There's lots so but pretty much the main thing that I've found it useful for is it gets rid of those annoying errors that kept happening. So it's a win.
- [05:46]
A
This really is. I mean there's so much really to call out here. First of all, just a quick note for everyone. You want to go to the AWS Labs GitHub repo. So this is not like on the AWS documentation, but it has its own separate GitHub repository and there's just like so much that you can be able to do with it. And of course for all our friends who have different preferences for their infrastructure as code, you'll be able to find a lot of different, different variations. Whether that's terraform, CDK cloudformation. I think that's, I think part that people really care about and then just even be able to access like the latest documentation, AWS pricing, cost explorer. So I think there's just so many different clever things that you can start to be able to do with this.
- [06:34]
B
Yeah, but 100 like yeah, if you use AWS run, don't walk and start using this, it will change. But remember to tell the LLM to use the Call AWS tool instead of the Use AWS tool. That's the little syntactic sugar that makes life easy.
- [06:55]
A
I like it.
- [06:56]
B
So Gillian, what are we talking about today? I've got a bunch of things to start with. I think you're going to start with analytics today.
- [07:01]
A
That's right. AWS Clean Rooms now supports collaboration with cross region data sources. This launch enables companies and their partners to easily collaborate with data sources that are stored in different regions without having to move, copy or Share the underlying data with clean room support for cross region collaboration. That is just going to make it a lot simpler, especially if you're thinking about being able to really optimize those costs. Cross region and another one from clean rooms. It now supports data access budgets this new privacy control allows you to limit the number of times your data can be analyzed when training or running inference on a custom machine learning model or in a SQL query or a pyspark job. So when you spend a budget, the system prevents additional analyses until the budget refreshes. But you can reset or edit a budget at any time as your needs change.
- [08:03]
B
And so this is not a cost budget, this is a data access budget, which is kind of cool.
- [08:07]
A
That's a really good call out. Yeah, a data access budget, not a cost budget. All right, next one is a glue adds write operations for SAP, OData, Adobe Marketo, Engage, Salesforce, Marketing, Cloud and HubSpot connectors. So this feature simplifies building end to end ETL pipelines that both extract data from and write process results based back to target applications. Amazon Kinesis Video streams now supports IPv6.
- [08:41]
B
For streams capability it is the year of IPv6. The year of IPv6.
- [08:49]
A
That's what we need to give out at like reinvent to people Shirts that.
- [08:52]
B
Say that Shirts that say that Amazon.
- [08:55]
A
OpenSearch ingestion now supports batch AI inference. Inference generates enrichments such as vector embeddings, predictions, translations and with these, real time inference that's ideal for low latency requirements. Now with OpenSearch ingestion you can use these same AR connectors with Amazon OpenSearch ingestion pipelines as an asynchronous batch inference job to enrich large data sets such as generating and ingesting up to billions of in vector embeddings.
- [09:27]
B
Let's talk application integration and we're happy to announce Apache Airflow 3.0 support in Amazon. Managed workflows for Apache Airflow so this is the latest major release of the workflow orchestration platform and this enhances your ability to author, schedule and monitor complex workflows with greater efficiency and control. So you get all the new cool stuff. There's a new scheduling system, a completely redesigned interface. Security and isolation has also been strengthened through the task execution API. There's also now scheduler managed backfill functionality as well. So for historical stuff also now supports Python 3.12 as well a whole bunch of security enhancements like it's it's a full release, it's a. It's a kind of a big deal. And AWS End User Messaging now sends onboarding progress alerts via Slack email or any other eventbridge destination. So before this launch, tracking the status of your onboarding process was difficult. You had to check the status of a phone number registration in the console. Now it will tell you at any stage when they're created, submitted, denied or required an update as well. So if you're not aware, AWS End user Messaging gives developers a scalable and cost effective messaging infrastructure without compromising the safety, security or results of their communications. This is also useful for things like one time passcodes, signups, account updates, appointment reminders, delivery notifications, promotions, and a whole bunch more.
- [10:58]
A
Next topic is artificial intelligence Amazon Bedrock now offers cohere embed version 4, the latest state of the art multimodal embedding model from Cohere that produces high quality embeddings for text, images and complex business documents. Traditional embedding models often struggle to understand complex multi bottle business materials. Now embed v4actually addresses this challenge by natively processing documents with tables, graphs, diagrams, code snippets and handwritten notes. The Open Source Model Context Protocol Server is now available for Amazon Bedrock Agent Core. This new standardized interface allows developers to analyze, transform and deploy production ready AI agents directly in their preferred development environment. You can use natural language to iteratively develop your agent, including transforming your agent logic to work with the Agent Core SDK and deploy your agent into development accounts. The open source MCP server is available globally via GitHub, so you're going to want to look this up in the Agent core MCP server GitHub repository, so that's just something to look out for. Amazon Bedrock Data Automation now supports enhanced transcription output for audio files by providing the option to distinguish between various speakers and separately process audio from each channel. This is definitely a good call out because I think a lot of people when they think about Bedrock Data Automation, they usually think about it specifically for extracting data, maybe from PDFs, but here's a great example of how you can actually use it with audio files. And in fact what they've done is they've expanded support for blueprint creation. So these are repeatable templates for certain types of problems using a guided and natural language based interface for extracting custom insights to audio modality. Speaker diarization detects each unique speaker and tracks speaker changes in a multi party audio conversation. Channel identification enables separate processing of audio from each channel. So one example could be let's say you've got Speakers such as a customer and a sales agent, they can be separated into unique channels and this can just make it easier for additional downstream processes that you might be having, depending on how you want to analyze. Maybe for the speaker saying versus what the other person on the other side is saying.
- [13:26]
B
That's very cool.
- [13:28]
A
Yeah, it is. So the AWS Knowledge MCP server is now generally available, which you can also find in the AWS API MCP that we were talking about earlier with this release. The server also includes knowledge about the regional availability of AWS APIs and Cloudformation resources.
- [13:46]
B
All right. That is so handy. Lots of people want that.
- [13:51]
A
Yes. Oh my gosh. And most importantly, it is publicly accessible at no cost. It does not require an AWS account.
- [14:01]
B
Fantastic.
- [14:03]
A
Amazon Q Developer now helps customers understand service prices and estimate workload costs. Customers can now use Q Developer to retrieve detailed product attribute and pricing information using natural language, making it easier to understand the cost of new workloads without having to review multiple pricing pages or specify detailed API request parameters. So you can literally type how much does RDS extended support cost? The cost of a plan workload or another example is like, I need to send a million notifications per month to email. How much is that going to cost? So I think this is super exciting for businesses of all different sizes.
- [14:46]
B
Yeah, this is.
- [14:47]
A
You're thinking about a new project and you're like, how much is this going to cost? And then before you start actually going, building it out. So that's really cool.
- [14:54]
B
Also have to make choices because you can say, hey, what would it cost if I did it this way? What would it cost if I do it that way? What are the pros and cons? All that sort of stuff. It's really cool.
- [15:01]
A
Totally, yeah. Really excited about that one. Amazon SageMaker Unified Studio announces corporate identity support for interactive Apache Spark sessions through AWS Identity Center's Trusted Identity Propagation. This new capability enables seamless single sign on and end to end data access, traceability for data analytics workflows. Amazon SageMaker Notebook instance now supports Amazon Linux 2023.
- [15:29]
B
Let's talk about the AWS Marketplace. If you've ever used the AWS Marketplace, it's actually a great place to get access to all kinds of software and APIs as well that are out there. I know for a lot of customers it's just a lot easier to just say, hey, I can just go to the Marketplace, get the thing I want, spin it up, and away we go. So a couple of things that have changed. The AWS Marketplace now supports new currency for usage based private offer, so it now supports Euro, pounds, Australian Dollars, Aussie, Aussie, Aussie and Japanese yen as well. So this means you can now reach buyers globally without currency conversion complexity or foreign exchange risk because you can price your offers in those currencies. And the AWS Marketplace has also announced enhanced pricing dimension capabilities for sellers. So this increases the maximum number of pricing dimensions from 24 to 200. So it means you can use new SaaS dimensions and you can also remove the 90 day price update restrictions for dimensions without active subscriptions. So this just means you can make things suit your customers better. Now speaking of suiting customers, let's talk business applications. The Amazon Connect team if I had to vote for the team that does the absolute highest rate of changes for their customers, it would have to be this team. I think we oh we always talk about stuff. They have a list of things I'm going to run through now. Like they've done a great job. So if you're not familiar with Amazon Connect, it is basically a cloud based contact center solution. Super popular, really has transformed a lot of organizations we've worked with, even locally as well because it just gives you a modern experience that works well both for your customers or citizens and also for the people taking the calls and processing the calls as well. So here's some of the things that have happened. Firstly, Amazon Connect now provides generative AI powered email conversation overviews and selected responses. So it enables the customer agents to handle emails more efficiently and it means customers get faster more consistent support. Amazon Connect now makes it easier to get customer input on outbound calls, so it now allows a prompt to be played to a customer on an outbound call after they answer the call but before they're connected with an agent. And Amazon Connect now enables you to customize service level calculations as well. So for example, managers can choose to count callback contacts or exclude contacts transferred out while waiting in a queue. Exclude short abandons using a configurable time threshold like lots of tweaks available there to measure stuff. It also now supports agent screen recording for Chrome OS devices, so with screen recording you can identify areas for agent coaching. And it's not just about listening to the customer call or reviewing the chat transcripts, but you can also watch the actions the agent is taking so you can find those efficiencies there as well. Amazon Connect now supports copy and bulk edit of agent scheduling configurations so you can do lots at one time. It also now supports agent schedule adherence notification as well. So you now can get notifications to supervisors when agents exceed their adherence thresholds. There's Also new case APIs to link related cases, add customer related items and search across them as well. And they Amazon Connect now also provides an Agent time off balance data in Analytics Data Lake so you can understand the balance of the latest and historical agent time off balances across different categories. So things like pay time off, sick leave, leave of absence, et cetera.
- [18:56]
A
I'm always impressed every single time. They do a lot they do Speaking of other things, let's talk about compute Amazon EKS and Amazon EKS distro now support Kubernetes version 1.34. Amazon ECS now supports one click event capture and event history querying in the AWS Management Console and we've got a new Compute optimized instance, the C8i and the C8i Flex instances. These instances are powered by custom Intel Xiyon 6 processors available only on AWS, delivering the highest performance and fastest memory bandwidth among the comparable intel processors in the cloud. The CAI and C8iflex instances offer up to 15% better price performance and 2.5x more memory bandwidth compared to previous generation intel based instances.
- [19:53]
B
So yeah, that's a big jump that one.
- [19:55]
A
It really is. So they deliver up to 20% better performance than C7I and C7I flex instances with even higher gains for some workloads. And they're up to 60% faster for Nginx web applications, up to 40% faster for AI deep learning recommendation models, and 35% faster for memed cache stores compared to C7i and C7i flex instances often used for web and application servers, databases, caches, Apache, Kafka, elasticsearch and Enterprise applications. And they're in a lot of different sizes, whether it's you need small going up to 16x large. Super excited for those. Amazon EC2 Instance Connect endpoint now supports IPv6 connectivity because it's the year of IPv6 now people are going to be sleeping and have that dream every time I see IPv6.
- [20:59]
B
Now this is what happens when you give me buttons to press.
- [21:07]
A
And we've got another new instance that's generally available, the M8A instances. These are powered by the 5th gen AMD EPYC processors formally codenamed Turn with a maximum frequency of 4.5 GHz and they deliver up to 30% higher performance and up to 19% better price performance compared to the M7A instances. The M8 instances deliver 45% more memory bandwidth compared to M7A instances, making these instances ideal for even latency sensitive workloads. They deliver even higher performance gains for specific benchmarks. 60% faster for Groovy JVM, up to 39% faster for Cassandra. And also compared to the M7A instances, they also are SAP.
- [22:01]
B
That's a big boost there. That's, that's a big like again, it reminds you that if you're using certain workloads, you've got to, you got to test these new instance types. Just, yeah, spin it up. Run a performance test. Like, wow, if I work up to 60% faster for my application, I'd be like this is awesome.
- [22:17]
A
It really is. And it's also just a reminder, I think for people who are using EC2 instances, you have to keep looking at what's new. And of course that's what we're here for to see. Hey, are there newer instances that I could be taking Advantage of? Correct EC2 image builder now automatically disables pipelines after consecutive failures and allows customers to configure custom log groups for image pipelines. AWIS Parallel Computing service now enables you to modify and update key Slurm workload manager settings without rebuilding your cluster. AWS Parallel Computing Service now supports node reboot via Slurm. AWS Parallel computing service expands Slurm's customization capabilities, enabling you to set over 60 additional parameters for granular control over your high performance computing cluster operations.
- [23:09]
B
That's a whole lot of slim going on.
- [23:12]
A
It really is. Next topic we've got is databases. Amazon Keyspaces for Apache. Cassandra now supports, you guessed it by.
- [23:22]
B
Now, IPv6, the year of IPv6.
- [23:29]
A
And just when you thought that was going to be our last time we talk about IPv6. Amazon DynamoDB now supports, you guessed it.
- [23:41]
B
Lots of buttons.
- [23:43]
A
Amazon Neptune now integrates with Graphstorm for scalable graph machine learning. Amazon relational database for DB2 now enables customers to perform native database level backups, offering greater flexibility in database management and migration.
- [24:02]
B
Let's talk developer tools. One quick but important update AWS Builder ID now supports Sign in with Google. So the AWS Builder ID is a personal profile that provides access to AWS applications, including Cairo, the AWS Builder Center, AWS training certification, AWS repost, and AWS startups. The AWS Builder ID represents you as an individual and is independent from any credentials and data you may have in existing AWS accounts. That's why it's important. And like other personal profiles, Adabus Builder ID remains with you as you progress to your personal educational and career goals. So it's always there for you, no matter who you're working for, what you're doing, et cetera. Now sign in with Google offers a convenient way for you to access those those capabilities with a single click using your Google account, which means you don't have to have separate credentials. If you'd like to use Sign in with Google, you can now let's talk about gaming. Amazon gamelift servers add the capability to view and connect to instances in the console. So if you like to have a look at stuff that way you can. And guess what. Amazon Gamelift Streams adds IPv6 support for streaming Windows applications. It just keeps on happening. Let's talk about management and governance. AWS introduces self service invoice correction feature to update AWS invoices. This is cool. So this launch enables all AWS customers to correct key invoice attributes, including purchase order numbers, business legal name and addresses on their AWS invoices and get corrected invoices instantaneously. This is massive. If you work in enterprise, if you do this sort of stuff, it just means that the whole cycle time is instant rather than having to wait for stuff going on. So I'm pretty excited about that. One we'd want to be excited about, but I know there's a lot of people for whom the quality of life improves. Speaking of quality of life, automatic quota management is now generally available for AWS service quotas. So this is a new capability, it's called automated quota management. And basically what AWS service quotas does, if you've never used it before, it helps you view and manage your quotas from a central location. Now this new feature monitors quota usage and notifies customers before they run out of allocated quotas supported on the AWS service quotas. This is awesome because it basically gives you a warning before you're going to hit the limit. And so you can configure your preferred network. So your preferred notification channels, I should say things like email, SMS or Slack through the console or through the API as well. And these notifications are also of course available in the Adbus Health page as well. So good way to keep on top of things. And the application map is now generally available for Amazon cloudwatch. So this lets you monitor large scale distributed applications by automatically discovering and organizing services into groups based on configurations and their relationships. I'm going to use this because I've got an application that fits this right now and I don't know what is going on with it, so I'm going to do this so with this new application Performance Monitoring APM capability, customers can quickly visualize which applications and dependencies to focus on while troubleshooting their distributed applications. That sounds handy. Quick update for Migration and Transfer AWS Data Sync now supports VPC Endpoint policies, so this allows you to control access to data sync APIs operations through the DataSync VPC service endpoints and FIPS 143 enabled VPC service endpoints. This is really useful depending on the security posture and compliance requirements that you have. Let's talk networking and content delivery. AWS Client VPN is Now supporting the macOS Tahoe release, so if you're on the latest and greatest, you can take advantage of that. And AWS Secrets Manager has expanded Private Link support to FIPS endpoints as well, so this is useful if you're in the FIPS world. And starting today, Amazon VPC Lattice lets you configure the number of IPv4 addresses assigned to Resource Gateway Elastic Network interfaces. So this enhancement builds on VPC Lattice's capability of providing access to resources on layer four things like databases, clusters, domain names, et cetera across multiple VPCs and accounts. So you can now specify the number of IPv4 addresses per ENI, which becomes.
- [28:23]
A
Immutable after setting now onto Security Identity and compliance. AWS Directory Service now enables customers to upgrade managed Microsoft AD from Standard to Enterprise Edition programmatically through the Update Directory Setup API. Amazon Cognito allows you to configure Terms of Use and privacy policy documents for managed login pages.
- [28:50]
B
That's actually really cool. I use the Manage login pages because quite frankly I can't be bothered building login pages. And it's always the sort of thing that you're just going to muck up, you're going to get it wrong. Cognito has a built in and this means you can customize it more. So big ups for Cognito Tank.
- [29:05]
A
And this was no code before everyone started not writing code by vitamin link.
- [29:09]
B
This is true.
- [29:12]
A
Amazon Detective now supports the Amazon VPC endpoints via Private Link, enabling you to securely initiate API calls to Detective from within your VPC without requiring Internet traversal. AWS Directory Service oh, of course we could not end the update show today with another IPv6. AWS directory service now supports, you guessed it, IPv6 connectivity from Manage, Microsoft Active Directory and AD Connector because it is.
- [29:44]
B
The year of IPv6. Indeed it is. I'm so glad we finished on that one. This is like an IPv6 heavy, heavy list. It's all here. It's all happening.
- [29:56]
A
I think people are going to be thinking about IPv6 way more than they have before after this episode.
- [30:02]
B
Yeah. Thanks, Simon. That's awesome. Well done AWS podcast. Thanks so much for helping. In all seriousness, I mean, IPv6 is critical for the future of the Internet, so the fact that it's all there is not a small thing. Whilst we like to make light of it, it just shows that it's happening, which is fantastic.
- [30:23]
A
Totally.
- [30:24]
B
So, Jillian, how do folks reach out to you?
- [30:26]
A
Jillian Ford on LinkedIn.
- [30:29]
B
And if you're old school, AWS podcast@Amazon.com is the place to do it. And until next time, keep on building.