Category: Daily Reading List

Another very good day on this European trip. We got to engage with 800+ developers in Istanbul, and then I met with a handful of local customers. I’m so proud of our team that planned and put on these events. And I’m also very much looking forward to flying home tomorrow.

[article]. AI Coding Assistants Are Reshaping Engineering — Not Replacing Engineers. We’re going to keep seeing articles like this. Nobody knows what’s going to happen, and folks will make some educated guesses.

[article] Google’s native multimodal AI image generation in Gemini 2.0 Flash impresses with fast edits, style transfers. Try the native image generation baked into Gemini 2.0 Flash. You can do it for free in Google AI Studio.

[blog] Agentic AI Is The Next Competitive Frontier. Don’t skip some investigation here, even if you’re not ready to invest.

[blog] Welcome Gemma 3: Google’s all new multimodal, multilingual, long context open LLM. Detailed, informative post from Hugging Face. As usual.

[blog] Mastra is moving into beta. I wouldn’t bet on Python being the dominant language for agents. Here’s a TypeScript framework, and you’ll see more coming out for a variety of languages.

[article] ServiceNow releases no-code, low-code AI agent builder. Forget Python, maybe the dominant agent-building language will be English?

[blog] AI IDEs Need Moats. There’s more hot-swapping of IDEs now than at any point I can recall. Folks are following the models to whatever buzzy IDE is available.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Arrived in Istanbul and I barely know what day or time it is. But, it’s been a good trip. I was happy to be in Paris for the Gemma Developer Day where we heard about our brand new open model. See more about it below.

[blog] Introducing Gemma 3: The most capable model you can run on a single GPU or TPU. We shipped the latest versions of our open LLM and it’s pretty hot. Scan the technical report too.

[blog] Introducing Gemma 3: The Developer Guide. Extra details on Gemma, our post-training strategy, and how to start using it.

[blog] How to deploy serverless AI with Gemma 3 on Cloud Run. You can try out Gemma on Google Cloud with a single CLI command, and also follow these docs for a more production-grade hosting on the same serverless stack.

[blog] Software Falsehoods: you can build it cheap, fast, and good – pick two. The point here is that “price” stands alone and isn’t a factor of speed or quality. Interesting insights here.

[blog] A 10x Faster TypeScript. The creator of C# wanted to improve TypeScript performance so he chose … Go. Good decision! Some .NET fans are salty about the choice.

[blog] Who gets to do strategy? Do engineers and other individual contributors have “permission” to do strategy? Will says everyone should be involved and empowered.

[blog] Gemini Robotics brings AI into the physical world. Wow, this is neat. Here’s a Gemini 2.0-based model that’s purpose built for robotic devices.

[blog] The Question Is No Longer “If” But “How” AI Is Transforming Software Development. Good extended take from Diego, who has been tracking the AI-assistance space longer than most.

[article] Checklist for Kubernetes in Production: Best Practices for SREs. Here’s a look at what cluster characteristics to account for, and even some pitfalls to avoid.

[blog] ScaNN for AlloyDB: The first PostgreSQL vector search index that works well from millions to billion of vectors. Some outstanding performance here, which matters with small or large vector stores.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Today was excellent as I got to spend time with lots of great developers at our Paris event, and then dinner with an extended team of Googlers. Tomorrow, off to Istanbul to meet more folks.

[blog] Debts, Tech and Otherwise. Ted looks at tech debt, the different types of debt, and why it’s sometimes a good choice to take on debt.

[blog] Agentic AI : Building a Multi Agent AI Travel Planner using Gemini LLM + Crew AI. I’m giving you a heads up that you’ll see more agent-related content in this reading list over the coming months. Mainly, because people are writing so much more about it!

[article] Leaders Can Move Fast and Fix Things. Earn the “right to run” and use speed responsibly to create change.

[blog] Building Websites With Lots of Little HTML Pages. Simon looks at this awkward acronym for an very interesting approach.

[blog] Developer-focused sessions and talks at Google Cloud Next ’25. Brad calls out some top talks—including one of mine—and other things that make this such a good event for developers.

[blog] What’s next after the AI bubble bursts? There’s a gold rush happening, but all these investments will stick around, as Sean points out.

[blog] Build Responsive, AI-powered Apps with Cloud Functions for Firebase. Streaming support for functions, among other new features in Firebase.

[blog] Benefits and Challenges of Infrastructure From Code. You know Infrastructure as Code, but what about Infrastructure from Code? Brian explains, with lots of links!

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Arrived in Paris yesterday and took myself to the Louvre museum to look around. Amazing. Today was filled with presentations and meetings, and prep for a talk tomorrow in one of the most interesting venues I’ve ever spoken in.

[blog] The State of LLM Reasoning Models. Deep thoughts in this piece on thinking models. Maybe more than you (or I) want to know, but it’s good background info to scan.

[blog] Model Context Protocol explained as simply as possible. Now I get it. Sean does a good job explaining what MCP is, and why folks are thinking it’s a big deal.

[blog] Model Context Protocol (MCP). Matt tackles MCP too. Longer piece, but also explains why abstracting tools (in addition to prompts and resources) is important for portability and agents use.

[blog] Guide: Our top four AI Hypercomputer use cases, reference architectures and tutorials. We talk about our “AI hypercomputer” but is it really a thing, or just an architectural concept? This post shows how you tangibly use it for AI workloads.

[blog] Implementing DevOps practices in front-end development. CI/CD, not DevOps per se, but whatever. Still a good reminder that every aspect of the stack should be testable.

[blog] Node just added TypeScript support. What does that mean for Deno? TypeScript support was a big part of the Deno differentiation from Node. Did that go away? Doesn’t seem like it.

[article] Vulnerability-Free Java Containers: A Practical Guide. Even if you have automation to constantly rebuild your container images to pull in the latest patches (good job!), you still depend on base images being maintained well! This post points out which have trouble keeping up.

[blog] Introducing built-in performance monitoring for Vertex AI Model Garden. Good feature. If you’re using one of our managed models (Gemini or others), we show you some at a glance info for usage and error rates.

[article] Generative AI adoption stymied by technology, talent woes. Survey after survey seems to show this mismatch between org goals and ability to execute on them.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

It’s the end of a good week for me, and tomorrow I’m off to Paris for events and customer meetings. I haven’t packed yet, because I’m a last-minute packer. You?

[blog] What Are Agentic Workflows? Patterns, Use Cases, Examples, and More. I liked this post as it does a good job explaining agents, and some important patterns.

[article] How a DevEx Initiative Aims To Save 500,000 Developer Hours. If you help each developer eek out a few extra hours of joyful coding, you’ll be in amazing shape.

[article] Accelerating engineering velocity at Block. More on the same story from above, with additional details. What projects are driving this new velocity for their developers?

[blog] Will the future of software development run on vibes? Are you vibe coding in your free time? If that makes no sense to you, read this.

[blog] Building AI Agents with Google’s Gen AI Toolbox and Neo4j Knowledge Graphs. I can imagine how useful graph databases would be to agents trying to research or plan work. Follow along with this example.

[blog] Great software design looks underwhelming. Important point here. Sophistication is often an anti-pattern.

[blog] Hallucinations in code are the least dangerous form of LLM mistakes. Another from Simon, but this feels like a very important point to amplify. Learn how to read and review code written by others, even LLMs!

[blog] Discovering new words with confidential federated analytics. Federated learning is cool, but I can imagine that folks can still feel weird about their device being part of the borg. This new technique should assuage some of those fears.

[blog] Gemini 2.0 Deep Dive: Code Execution. It’s exciting to see LLMs able to generate *and* execute code via API. Follow along here using Google AI Studio too.

[blog] Code Reviews with AI: a Developer Guide. AI shouldn’t just help you crank out code; it should be helping you improve quality as well.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

I’ve got some things for you today that may cause angst—”why are colleagues jealous of me?”, “gah, our security situation is getting worse!”, “wait, I need to become a great storyteller?”—but it’s good to lean into hard problems.

[blog] Applied “Software Engineering at Google.” Wow, terrific piece by Addy. These are practices that anyone should consider adopting, even if you’re not Google-scale.

[article] Automation Can Solve Resource Overprovisioning in Kubernetes. Wow, virtually every Kubernetes cluster in the cloud has too much CPU allocated? There’s some advice here to fix it. More coverage on this topic here.

[blog] How I Code With LLMs These Days. Each of us is figuring out how these new tools work for us. Here’s one take.

[blog] Hej Sverige! Google Cloud launches new region in Sweden. The global expansion of Google Cloud keeps going. I’m happy that our Swedish friends now get access to the best!

[blog] AlloyDB Omni on Mac with Podman. This is a demo scenario, but you can run AlloyDB Omni for production use cases on any infrastructure. Pretty cool.

[article] When Your Colleague Is Jealous of Your Success. The world’s not driven by greed; it’s driven by envy. It’s a dangerous mindset, and we need to be aware of it.

[blog] Every Developer Deserves Great Documentation. They sure do. What “great” looks like is debatable, but you know it when you see it.

[article] Software Development Teams Struggle as Security Debt Reaches Critical Levels. Yeesh, not good. It’s taking teams longer to roll out security fixes to their code and systems. This might not get better as we generate even more code.

[blog] Not just for developers: How product and security teams can use GitHub Copilot. AI isn’t just for the IDE; all sorts of folks can use it on all sorts of surfaces to get their job done.

[blog] Why everyone can (and should) be a great storyteller. Heck yes. This matters more now than ever, and these skills will differentiate you in an AI-saturated market.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

It’s happening. Day by day, I find myself using more AI stuff to do my daily job. Today it was help generating ideas and example code for a library that seemed complicated to start with from scratch. And then doing two different “research” efforts for things I’m writing about. You resisting, or starting to incorporate AI more?

[article] Is Incrementalism Holding Back Your AI Strategy? Maybe you’re playing it safe and aiming low with AI. Just update this current process, or tweak that product. This article encourages you to recognize the high stakes.

[blog] Function Calling Guide: Google DeepMind Gemini 2.0 Flash. So good. Philipp digs into what function calling is all about and how to do it. This is such an important topic for those building AI apps.

[article] “I scratched my own itch” isn’t good enough. Excellent reminder to get outside and talk to people. Just because your product idea was brilliant for solving your problem doesn’t mean it’s relevant to others.

[article] People are using Super Mario to benchmark AI now. This is the way. Real intelligence is knowing all the hidden rewards and warp zones in this game.

[blog] Ensuring AI Code Quality with SonarQube + Gemini Code Assist. There probably won’t be one product to rule them all. Plugins, components, and agents will work together to make our software and systems better.

[blog] Do developers need code samples in API documentation? My team owns docs and code samples, so a discussion like this piques my interest. What do people want out of code samples in docs nowadays?

[blog] Streamline your LangChain deployments with LangServe. You’re going to see a lot about deploying and hosting agents. The tech is emerging quickly. Here’s an example of a workload you can containerize and run in Kubernetes.

[blog] Expanding AI Overviews and introducing AI Mode. Our first AI experiments in Search were well-intentioned but a bit clumsy. It’s looker better now.

[article] How do developers fix bugs differently when they don’t author them? We fix our own bugs faster, which might not be a surprise?

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Today was a great day. Fourteen meetings, but still great. I’m getting prepped for a trip to Paris and Istanbul next week, and feeling grateful that I have an enjoyable job working alongside terrific folks.

[article] If Architectural Experimentation Is So Great, Why Aren’t You Doing It? Are they great? You need to answer that first. But assuming “yes”, this article explains why they’re hard to sell.

[article] How to Get Better at Delegating. Super important skill. Even if you’re an individual contributor, you have opportunities to delegate tasks to other humans, or even AI.

[blog] Evaluate gen AI models with Vertex AI evaluation service and LLM comparator. Do this with us, do this with someone else. Just do it. Have a system in place to evaluate your models!

[article] The TechCrunch AI glossary. Get some definitions for terms such as “deep learning” and “chain of thought.”

[blog] Beyond the Debugger: A Comprehensive Guide to Debugging Go Applications. This is written for Go developers, but virtually all of the advice applies to any other programming language too.

[blog] LLMs.txt to help LLMs grok your content. I learned a few things here, and found it super cool that Guillaume could analyze all his blog posts in a single shot to Gemini.

[blog] How to calculate your AI costs on Google Cloud. “What does it cost?” That’s more easily discoverable in the cloud than on-premises, but still requires some investigation.

[blog] Platform Building Antipatterns: Slow, Low, and Just for Show. If you’re doing any of these things, there’s still time to stop, reassess, and make positive changes.

[article] Salesforce’s AgentExchange launches with 200+ partners to automate your boring work tasks. You’ll start seeing agents everywhere. They’ll be of mixed quality depending on data sources, underlying models, and skill of the developer who built them.

[article] Why AI Agents Suck So Bad. Some do, sure. If they’re just glorified database lookups, you won’t get the nuanced responses needed for complex situations.

[article] 5 content marketing trends to keep an eye on in 2025. I found this to be a good list of places where you could revisit your marketing strategy, especially using AI.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Happy Monday. It’s March already, and today’s list has a lot of content that will make you smarter. Or you already knew all this, and I’m the only one who didn’t.

[article] Orchestrate Cloud Native Workloads With Kro and Kubernetes. A few weeks back, I mentioned this open source tool created by a partnership among cloud vendors. Jani takes it for a spin with a real example.

[article] Cloud Giants Collaborate on New Kubernetes Resource Management Tool. Related to the above article. Here’s additional color on this tool.

[article] Pioneering the Future: Advancing Infrastructure for AI Agents. I’ve seen lots of talk about agents, much less about the infrastructure needed to support them.

[article] Google upgrades Colab with an AI agent tool. This Data Science Agent looks ready to go. Seems like a useful way to ask questions of your raw data.

[blog] Weak managers. This spoke to me. I need a manager with clout so that my team can get things done, and I try to be that type of manager myself.

[blog] Perspectives on Generative AI in Software Engineering and Acquisition. The Software Engineering Institute looks at what it takes for generative AI to be useful and trusted for various tasks.

[blog] Every pod eviction in Kubernetes, explained. Great level of details here from Ahmet. This might be something you don’t want to think much about, but you should be aware of it.

[article] Google Cloud Becomes World’s #1 AI Partner via Landmark Salesforce Alliance. Simmer down, Bob. I’m happy about it too. But we all have a long way to go!

[article] The 4 Types of Thinking Leaders Need to Practice—and Teach. I have an untested belief that too many (tech) leaders spend too little time thinking. Instead of so much operational and tactical focus, I’d love to see more people create the space to truly think.

[blog] Multi-Tenant: Database Per Tenant or Shared? I haven’t seen many platforms that make it easier to create and deliver SaaS platforms. Feels very DIY? Derek looks at one fundamental choice you have to make.

[article] Rolling out developer productivity metrics. This is about more than the numbers. Abi explains the change management and traps that can pop up.

[article] Buy an expensive “AI Gateway”? Thanks, we’ll just build and open-source it, says Bloomberg. Do you have an AI gateway yet? Maybe you can use OSS instead of buying something right now.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:

Today’s reading list was a good reminder that even stable things will change. Hash tables in a programming language? That’s pretty standard stuff, and yet apparently still disrupt-able. And Skype? It feels like it’s always just been around. Not anymore.

[blog] Faster Go maps with Swiss Tables. This post has everything. There’s a history lesson, explanation of a topic in a way that normal humans can understand, and a good payoff. Learn how Go changed built-in map data type based on an improved hash table design.

[blog] Non-Infrastructure-as-Code-based Cloud and Kubernetes management and automation tools. There’s a whole world of Kubernetes tools beyond IaC systems. Brian does an audit of what’s out there.

[blog] AI APIs in 2025. Here’s an overview (with examples) of some top LLM-as-a-service providers and endpoints.

[blog] Inter-VPC connectivity architecture patterns in Cross-Cloud Network. If you don’t need to set up a cross-cloud network, don’t do it. But nowadays, it’s relatively common, and I like that we document patterns like these.

[article] Google’s principles for measuring developer productivity. Abi takes a look at a new paper from our developer insights team. These seem like good principles to follow.

[article] End of an era: Microsoft to shut down Skype, shifting users to Teams, 14 years after $8.5B deal. I’ll need to choose a new way to video chat with my parents each Saturday (note: it will not involve Microsoft Teams), but it’s the end of an era. Skype was big time for a while!

[blog] FacetController: How we made infrastructure changes at Lyft simple. Interesting approach to standardizing (and simplifying?) deployments on Kubernetes.

[blog] Get Google Cloud certified in 2025—and see why the latest research says it matters. Certification still matters. Even if it’s just to force you to go deep on a topic and subject yourself to testing.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below: