About this episode
Welcome to episode 298 of The Cloud Pod – where the forecast is always cloudy! Justin, Matthew and Ryan are in the house (and still very much missing Jonathan) to bring you a jam packed show this week, with news from Beijing to Virginia! Did you know Virginia was in the US? Amazon definitely wants you to know that. We’ve got updates from BigQuery Git Support and their new collab tools, plus all the AI updates you were hoping you’d miss. Tune in now! Titles we almost went with this week: The Cloud Pod now Recorded from Planet Earth ☕Wait Java still exists? When will java just be coffee and not software Cloudflare Makes AI beat Mazes Replacing native mobile things with mobile web apps won’t fix your problems AWS Turn your security over to the bots The Cloud Pod is lost in the AI labyrinth AI security agents to secure the AI… wait recursion Durable + Stateless.. I don’t know if you know what those words means Click ops expands to our phones yay! The Cloud Pod is now a data analyst ⁉️Gitops come to bigquery A big thanks to this week’s sponsor: We’re sponsorless! Want to get your brand, company, or service in front of a very enthusiastic group of cloud news seekers? You’ve come to the right place! Send us an email or hit us up on our slack channel for more info. AI Is Going Great – Or How ML Makes All Its Money 00:46 Manus, a New AI Agent From China is Going Viral—And Raising Big Questions Manus is being described as “the first true autonomous AI agent” from China, capable of completing weeks of professional work in hours. Developed by a team called Butterfly Effect with offices in Beijing and Wuhan, Manus functions as a truly autonomous agent that independently analyzes, plans, and executes complex tasks. The system uses a multi-agent architecture powered by several distinct AI models, including Anthropic’s Claude 3.5 Sonnet and fine-tuned versions of Alibaba’s Qwen . Unlike traditional chatbots, Manus can work on different tasks without needing frequent, step-by-step instructions, continuing to work in the background even when users close their computers A unique feature is the “Manus’s Computer” window, which allows users to observe what the agent is doing and intervene at any point. The company claims Manus outperforms OpenAI’s Deep Research tool on the GAIA benchmark , a third-party measure of general AI assistants Early testing has shown mixed results – while some reviewers were impressed, others encountered bugs, error messages, and failures on practical tasks like ordering food or booking flights. The system remains difficult to access due to limited server capacity, creating a scramble for invitation codes which are reportedly selling for thousands of dollars on Chinese reseller apps. Manus has announced a strategic partnership with Alibaba’s Qwen team to help deal with the surge in traffic and expand its user base The emergence of Manus is raising questions about the global AI landscape, with some comparing it to January’s “DeepSeek moment” and questioning whether China has leapfrogged the US in AI development. Privacy experts have raised concerns about data protection, noting uncertainty about server locations and potential data transfers to China. 02:16 Matthew – “It’s no different than giving all your personal information to ChatGPT. Sure, I don’t want to give it to China. But I also don’t like giving it to OpenAI either. 04:14 Cloudflare turns AI against itself with endless maze of irrelevant facts – Ars Technica Cloudflare has announced a new feature called “ AI Labyrinth ” that aims to combat unauthorized AI data scraping by serving fake AI-generated content to bots. The tool will attempt to thwart AI companies that crawl websites without permission to collect training data for LLM that power AI assistants like ChatGPT . Instead of simply blocking the bots, Cloudflare’s new system lures them into a maze of realistic looking but irrelevant pages, wasting the crawlers computing resources. The approach is a notable shift from the standard block-and-defend strategy used by most website protection services. Cloudflare says blocking bots sometimes backfires because it alerts the crawlers operators that they’ve been detected. When Cloudflare detects unauthorized crawling, rather than blocking the request, it will link the bot to a series of AI-generated pages that are convincing enough to entice a crawler to traverse them. But while real looking, the content is not actually the content of the site they are protecting, so the crawler wastes time and resources. The data is automatically generated by its Workers AI service, a commercial platform that runs AI tasks. 05:40 Ryan – “Yeah, is the hallucination in the model? Or is it the bad data that it’s being fed?” 07:05 Introducing 4o Image Generation | OpenAI OpenAI has long believed image generation should be a primary capability of their language models. That’s why they have built “the most advanced image generator yet” into GPT-4o, the result image generation that is beautiful, but also useful. For example: 11:39 Introducing next-generation audio models in the API Open AI is launching a new speech to text and text to speech audio model in the API making it possible to build more powerful, customizable, and intelligent voice agents that offer real value. The latest speech to text models set a new state of the art benchmark, outperforming existing solutions in accuracy and reliability — especially when dealing with accents, noisy environments and varying speech speeds. These enhancements are in the gpt-4o-transcribe and gpt-4o-mini-transcribe models with improvements to word error rate and better language recognition and accuracy, compared to the original whisper models. Show note editor aside: As a historian (who specialized in Byzantine and early Medieval studies) tech jargon can sometimes be difficult for me to interpret just by ear. I can sometimes tell when the transcript is off, but sometimes I can’t, and more efficient transcripts would be awesome. Cloud Tools 12:44 Valkey 8.1’s Performance Gains Disrupt In-Memory Databases This article on Valkey caught Justin’s eye, as it’s been a year since Redis announced they were dumping the BSD 3-clause licenses and adopting the RSALv2 and SSPLv1 licenses. This is the event that birthed the Valkey fork . Apparently the Valkey fork is turning out to be highly successful, per a Percona research paper, 75% of surveyed Redis users are considering migration due to recent licensing changes, and of those considering migration 75% are already testing, considering or adopted valkey. Third party Redis Developer companies like Redisson are supporting both Redis and Valkey. It’s not just the licensing that’s driving, but at the Linux Foundation Member Summit , said that Valkey is far faster thanks to incorporating enhanced multi-threading and scalability features. That wasn’t the original plan , as they wanted to keep the open source spirit, but also wanted the value to be more than just a fork. Initially at the first contributor summit in Seattle where we got together developers and users to try to figure out what this new project would look like. At the time it was expected to focus on caching, but users said they wanted more, with Valkey being a high performance database of all sorts of distributed workloads, and although that would cause a lot of complexity, the new core team took that on. They were successful with Valkey 8 redesigning Redis’s single threaded event loop threading model with a more sophisticated multithreaded approach to I/O Operations which resulted in a 3x improvement in performance as well as 20% reduction in the size of separate cache tables. Beyond that they have been improving the core by adding rust to add memory safety. As well as changing internal algorithms to improve reliability and failover times. As well as they have rebuilt the key-value store from scratch to take better advantage of modern hardware based on work done at Google. A ton of this will come out as part of Valkey 8.1 . 16:18 Matthew – “The performance improvements here are massive…it’s pretty amazing what they’re able to do now.” If they keep improving, Redis is just going to slowly die off due to their own causes.” AWS 17:49 Detailed geographic information for all AWS Regions and Availability Zones is now available | AWS News Blog Starting today, you can get more “granular” visibility of geography. Amazon says that due to data sovereignty, the need for more details is super important. They have added Geography to the AWS Region and Availability Zones . Virginia is in the United States of America, in case you didn’t know. 21:22 Matthew – “So maybe FanDuel didn’t know that US East-1 is in Virginia, and in Virginia they can’t do gambling? So they got a fine there, but they can do it in Ohio, so now they know US East-2 is in Ohio.” Listener note: Is this update important to you? We’d love to hear more about that! Slack, X, Bluesky…you know where to find us. 22:33 New Capability of Amazon Q in QuickSight Makes Every Employee Their Own Data Analyst AWS has announced that Amazon Q in QuickSight unlocks the ability for any employee to perform expert-level data analysis using natural language, without the need for specialized skills or expertise. “We are at the beginning of a workplace transformation driven by agents, and Amazon QuickSight is pioneering how this technology can break down the technical barriers between employees and their data,” said Dilip Kumar, vice president of Amazon Q Business, AWS. “With the new scenarios capability, everyone becomes their data analyst who can dive deep into their company data, helping them unlock insights, make better decisions, and explore countless possibilities faster than ever.” 25:07 AWS announces OR2 and OM2 instances for Amazon OpenSearch Service Amazon Opensearch service introduces new instances of OR2 and OM2, expanding the opensearch optimized instance family. The OR2 delivers up to 26% higher indexing throughput than previous OR1 instances and 70% over R7g instances. The new OM2 instances provide up to 15% higher indexing throughput compared to OR1 instances and 66% over m7g instances in internal benchmarks. 25:27 Ryan – “It’s funny to see these announcements, years after running a giant Elasticsearch project for awhile. These are all the struggles, and they’re getting addressed through OpenSearch and Amazon running a giant farm of these things.” 26:42 Amazon Corretto 24 is now generally available Correto 24 has been released, which is the OpenJDK 24 feature release. The next LTS version will be Java SE 25, which comes out in September. The current LTS is 21. Considering everyone (including Justin) is still on Java 8, it might be time to upgrade. 28:59 AWS announces expanded service support in the AWS Console Mobile App If you are eternally disappointed in the AWS Mobile app and its limited coverage, the latest update might make you much happier: 24 additional services are now available including Service Quotas, Cloudfront, SES, Cloud 9, and AWS Batch via an integrated mobile web browser experience in the Console mobile app. Justin appreciates the effort – but mostly we’re just hoping they’re not abandoning mobile native completely for the mobile app. 33:32 AWS Network Firewall introduces new flow management feature AWS is giving you a new flow management feature for AWS Network Firewall that enables customers to identify and control active network flows. This feature introduces two key functions: Flow capture – which allows point in time snapshots of active flows Flow Flush, which enables selective termination of specific connections. 33:53 Justin – “ So flow capture is just the networking team is sick of providing packet captures, I imagine. So now it’s self-service. makes perfect sense.” GCP 33:04 Google Next is coming up in a few short weeks. Want to see Justin in person? And maybe even get some stickers? Check out these critical sessions: –BRK2-024 – Workload-optimized data protection for mission-critical enterprise apps –BRK1-028 – Unlock value for your workloads: Microsoft, Oracle, OpenShift and more 37:04 Introducing protection summary, a new Google Cloud Backup and DR feature Data protection is critical to your cloud strategy, and that includes backups and DR . Making sure your backups are set up correctly and aligned with your RPO/RTO requirements is critical. However, collecting the data in your complex cloud environment can be tricky. So Google is giving you a preview of the Protection Summary and the data protection tab, a new feature in Google Cloud Backup and DR that provides a centralized view of your backup configurations, helps you identify gaps in your data protection, and empowers you to take action to improve your resiliency. Protection summary will quickly help you identify resources with no backup configuration. Quickly configure backups for resources and then assess the backup configurations and vulnerability to ransomware. 38:25 Ryan – “ That was the first thing I was thinking about when I read through this was the the terrible-ness that I did 12 years ago to plug in some sort of backup errors to a slack channel so that we could pass an audit for notifications. It was ridiculous.” 39:23 Expanding Gen AI Toolbox for Databases with Hypermode Google recently announced the public beta of AI Toolbox for Databases , and today they are excited to expand its capabilities through a new partnership with Hypermode . Gen AI Toolbox for Databases is an open source server that empowers application developers to connect production-grade, agent-based generative AI (gen AI) applications to databases. Toolbox streamlines the creation, deployment and management of sophisticated gen AI tools capable of querying databases with secure access, robust observability, scalability and comprehensive management. Currently, the toolbox supports AlloyDB , Spanner , Cloud SQL for PostgreSQL, MySQL, and SQL Server , as well as self-managed MySQL and PostgreSQL. Justin doesn’t know what Hypermode is, so this announcement isn’t for him. But if you do know what Hypermode is, then today is a good day! 41:42 Announcing BigQuery repositories: Git-based collaboration in BigQuery Studio Modern data teams use Git to collaborate effectively and adopt software engineering best practices for managing their data pipeline and analytics code. But most tools don’t offer integration with Git version control systems, making Git workflow feel out of reach. This forces users to copy and paste code between UIs, which is not only time-consuming but also error prone. To help, they’re releasing in preview “ BigQuery Repositories ” a new experience in bigquery studio that helps data teams collaborate on code stored in git repositories. BigQuery repos provide a comprehensive set of features to integrate Git workflows directly into your BigQuery environment: Setup new repos in BigQuery Studio where you can develop SQL queries, Notebooks, data preparation, data canvases, or text files with any file extension. Connect your repositories to remote git hosts like GitHub, GitLab, and other popular Git platforms. Edit the code in your repositories within a dedicated workspace, on your own copy of the code, before publishing changes to branches Perform most Git operations with a user-friendly interface that lets you inspect differences, commit changes, push updates, and create pull requests all within BigQuery Studio. 46:06 Gemini 2.5: Our most intelligent AI model Google has introduced Gemini 2.5 , their most intelligent AI model. The first 2.5 release is an experimental version of 2.5 Pro, which is state-of-the-art on a wide range of benchmarks and debuts at #1 on LMArena by a significant margin. Gemini 2.5 models are thinking models, capable of reasoning through their thoughts before responding, resulting in enhanced performance and improved accuracy. With Gemini 2.5, Google has achieved a new level of performance by combining a significantly enhanced base model with improved post-training. Going forward, they will build thinking capabilities directly into all models, so they can handle more complex problems and support even more capable, context-aware agents. Google is very proud that 2.5 Pro takes the top of LMArena leaderboard Gemini 2.5 without test time techniques, like Majority voting, 2.5 leads in math and science benchmarks. It also scores a state-of-the-art 18.8% across models without tools used on Humanity’s last exam, a dataset designed by hundreds of SME to capture the human frontier of knowledge and reasoning. 2.5 will have a big leap over 2.0 on coding performance, as well as it excels at creating visually compelling web apps and agentic code applications, along with code transformation and editing. 47:27 Ryan – “Well, 2.o was a big fix over 1.5, so I’m hoping that it’s as big of an impact.” Azure 49:23 Announcing the public preview launch of Azure Functions durable task scheduler Microsoft is announcing the public preview of Azure Functions Durable Task Scheduler. This new Azure-managed backend is designed to provide high performance, improve reliability, reduce operational overhead, and simplify monitoring your stateful orchestrations. Durable functions provide you a simplified way to develop complex, stateful and long-running apps in the serverless environment. This allows developers to orchestrate multiple function calls without having to handle fault tolerance. It’s great for scenarios like orchestrating multiple agents, distributed transactions, big data processing, batch processing like ETL (Extract, Transform and Load), Async APis, and essentially any scenario that requires chaining function calls with state persistence. 47:27 Matthew – “I t’s step functions with a CloudWatch event that triggers it…It’s going to do everything that step functions can do.” 52:29 Announcing GA for Azure Container Apps Serverless GPUs | Microsoft Community Hub Azure Container Apps Serverless GPU’s are now GA. This allows you to seamlessly run your AI workloads on-demand with automatic scaling, optimized cold strat, per-second billing, and reduced operational overhead. Nvidia powers the serverless GPU’s which allows you to seamlessly run billing with scale down to zero when not in use. Thus, reducing operational overhead to support easy real-time custom model inference and other GPU-accelerated workloads. In addition this supports NVIDIA NIM microservices , which are part of the Nvidia AI Enterprise , its a set of easy to use microservices designed for secure, reliable deployment of high-performance AI model inference at scale. Key Benefits for Serverless GPU’s Scale-to zero GPUs: Support for serverless scaling of NVIDIA A100 and T4 GPUs. Per-second billing: Pay only for the GPU compute you use. Built-in data governance: Your data never leaves the container boundary. Flexible compute options: Choose between NVIDIA A100 and T4 GPUs. Middle-layer for AI development: Bring your own model on a managed, serverless compute platform and easily run your AI applications alongside your existing apps. 47:27 Ryan – “I want to make fun of this, but I love the fact that it scales to zero. If I were making some sort of application, I’d go bankrupt without something like this in place, so I think it’s kind of neat.” 54:53 Microsoft and NVIDIA accelerate AI development and performance Accelerating agentic workflows with Azure AI Foundry, NVIDIA NIM, and NVIDIA AgentIQ Microsoft and NVIDIA have several enhancements to help shape the future of AI. This includes integrating the newest Blackwell platform on Azure AI, incorporating NVIDIA NIM microservices into Azure AI Foundry , and empowering developers to accelerate their innovations and solve challenging problems. NIM provides optimized containers for more than two dozen popular foundation models, allowing developers to deploy generative AI applications and agents quickly. These new integrations can accelerate inference workloads for models available on Azure, providing significant performance improvements, greatly supporting the growing use of AI agents. Key features include optimized model throughput for NVIDIA accelerated computing platforms, prebuilt microservices deployable anywhere and enhanced accuracy for specific use cases. General availability of GB200 V6 virtual machine series accelerated by NVIDIA GB200 NVL72 and NVIDIA Quantum Infiniband networking. Once you have NVIDIA NIM deployed, Nvidia AgentIQ takes center stage with its open source toolkit designed to seamlessly connect, profile and optimize teams of AI agents, enabling your systems to run at peak performance. AgentIQ delivers: Profiling and optimization Dynamic inference enhancements Integration with Semantic Kernel 55:50 Justin – “ It gives you the PyTorch type tools, all the different capabilities you might want to use to use your GPUs effectively, to do training or inference – all prebuilt into the NIM containers that are prebuilt for you. That’s what it is. They made it sound like it was special, but it’s not.” 58:08 Microsoft unveils Microsoft Security Copilot agents and new protections for AI Last year Microsoft launched Security Copilot to empower defenders to detect, investigate and respond to security incidents swiftly and accurately. Now they are announcing Security Copilot with AI agents designed to autonomously assist with critical areas such as phishing, data security and identity management. The relentless pace and complexity of cyberattacks have surpassed human capacity and establishing AI agents is a necessity for modern security. Microsoft’s Threat Intelligence now processes 84 trillion signals per day, revealing the exponential growth in cyberattacks. Today, they are launching 6 Security Copilot agents built by Microsoft and 5 built by their partners available in preview in April. The five agents from Microsoft: The Phishing Triage Agent in Microsoft Defender triages phishing alerts accurately to identify real cyber threats and false alarms. It provides easy-to-understand explanations for its decisions and improves detection based on admin feedback. Alert Triage Agents in Microsoft Purview triage data loss prevention and insider risk alerts, prioritize critical incidents, and continuously improve accuracy based on admin feedback. Conditional Access Optimization Agent in Microsoft Entra monitors for new users or apps not covered by existing policies, identifies necessary updates to close security gaps, and recommends quick fixes for identity teams to apply with a single click. Vulnerability Remediation Agent in Microsoft Intune monitors and prioritizes vulnerabilities and remediation tasks to address app and policy configuration issues and expedites Windows OS patches with admin approval. Threat Intelligence Briefing Agent in Security Copilot automatically curates relevant and timely threat intelligence based on an organization’s unique attributes and cyber threat exposure. The five agentic solutions from partners include: Privacy Breach Response Agent by OneTrust analyzes data breaches to generate guidance for the privacy team on how to meet regulatory requirements. Network Supervisor Agent by Aviatrix performs root cause analysis and summarizes issues related to VPN, gateway, or Site2Cloud connection outages and failures. SecOps Tooling Agent by BlueVoyant assesses a security operations center (SOC) and state of controls to make recommendations that help optimize security operations and improve controls, efficacy, and compliance. Alert Triage Agent by Tanium provides analysts with the necessary context to quickly and confidently make decisions on each alert. Task Optimizer Agent by Fletch helps organizations forecast and prioritize the most critical cyberthreat alerts to reduce alert fatigue and improve security. 59:42 Ryan – “ So as the new security guy who’s learning all these tools and going through all the things that are in Microsoft Defender, I am very skeptical that this is going to actually solve any issues. But sweet Jesus, if it’s an improvement on what Microsoft Defender already does, it’d be welcome. The patterns and stuff that are detected natively in those tools just by default is not good enough, and so you have to spend a ton of time trolling through too much data to make these things work for anything other than forensic investigation after the fact.” Oracle 1:03:02 Oracle Introduces AI Agent Studio Oracle has announced Oracle AI Agent studio for Fusion Applications , a comprehensive platform for creating, extending, deploying and managing AI agents and agent teams across your enterprise. This is part of the Oracle Fusion Cloud Application Suite, the new AI Agent Studio provides easy-to-use tools for customers and partners to create customized AI agents that address complex business needs and can help drive new levels of productivity. Oracle AI agent Studio includes: Agent Template Libraries Agent Team Orchestration Agent Extensibility Choice of LLMs Native Fusion Integration Third-party system integration Trust and Security framework Validation and testing tools 1:03:41 Matthew – “Oracle showed up to the AI Agent party.” Closing And that is the week in the cloud! Visit our website, the home of the Cloud Pod where you can join our newsletter, slack team, send feedback or ask questions at theCloud Pod.net or tweet at us with hashtag #theCloudPod