Category: Daily Reading List

Did you have a solid week? Our industry is still bonkers with companies like Block resetting after over-hiring and figuring out what to expect from AI. All this while new tech floods our sense every day. Deep breaths. It’s a lot. I’m going to take time this weekend to watch my kiddo play Little League baseball and avoid dwelling on things I can’t control.

[blog] The Generative AI Policy Landscape in Open Source. How do 60 different open source organizations approach generative AI? Kate did the work, and the findings are very interesting.

[blog] Nano Banana 2: Combining Pro capabilities with lightning-fast speed. Impressive stuff from our best image model. Game changer for enterprise users too.

[blog] Signal Forms: Angular’s best quality of life update in years. You might not care about Angular, but this is still a good reminder that the ergonomics of your product/tool/framework make a big impact on people.

[blog] Skills Made Easy with Google Antigravity and Gemini CLI. This is a helpful post for those who are trying to install and manage skills across AI tools.

[article] Software vulnerabilities are being weaponized faster than ever. A quick reminder to have rapid responses in place, especially for software you control yourself.

[blog] Two Beliefs About Coding Agents. I second these. There’s a lot of unstated smarts going into the best prompts, and many of the AI “apps” are personal projects.

[article] I’m a Google exec who spends 20+ hours a week experimenting with AI. This is the best era to be a developer. Behind a paywall, I think. But a good story about my boss who uses AI every day for real work.

[article] Free Skate. Great reminders about the reality of pressure, and our choice in how we deal with it.

[blog] Serving data from Iceberg lakehouses fast and fresh with Spanner columnar engine. What if you didn’t need to do ETL to get data from your transactional database into a lakehouse for analytics? That possibility is now a reality.

[article] Enterprise MCP adoption is outpacing security controls. Yes, there’s a lot of surface area for problems here in the enterprise.

[blog] Securing AI Agents When Using Google Managed MCP Servers: A Defense-in-Depth Guide. Here’s one way to start getting proactive with your remote managed MCP servers. I like this advice.

[article] Google’s Opal just quietly showed enterprise teams the new blueprint for building AI agents. I need to take a second look at this Labs project from us. Seems like a powerful update.

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

I open the Gemini CLI at least a couple times a day, and it seems I have a waiting update every time. This thing gets refreshed constantly! I can’t tell if I love that or want us to slow down 🙂

[article] Packaging Expertise: How Claude Skills Turn Judgment into Artifacts. “Expertise transfer” is a smart way to label Agent Skills. That said, we’ll need to see how good we get at keeping these up to date; otherwise they atrophy like other doc sources.

[blog] Easily Build a Local MCP Server in Java with a Skill inside Gemini CLI. My “skills” FOMO is off the charts right now. Need to take time to build a few before the week is out. Guillaume does a good job showing how to use one to build MCP servers.

[article] The engineeringification of everything. It’s not just you; engineering is everywhere and expanding to all types of roles.

[article] Humans Are for Ideas, AI Is for Execution. I mean, I use AI to help me think through ideas, but the point here is valid. We’re the source of real thought!

[blog] Building a Memory-Augmented Supply Chain Agent with ADK, AlloyDB and Memory Bank. Agents with memory. They’re important. Read this post from Abi for a sense of the tiers of memory and how to bring them into your agents.

[blog] 5 More Things I’d Tell Myself on Day 1 of using Cloud Run. These are so great. Sara calls out a few things I didn’t know about, and I’ve been using Cloud Run for years.

[blog] What’s so hard about continuous learning? This is for models, not people. Could we keep running new input through a training pipeline? Yes. But there’s risk and quality concerns with doing that.

[blog] New in Confluent Intelligence: A2A, Multivariate Anomaly Detection, Vector Search for Cosmos DB, Amazon S3 Vectors, and More. Cool to see enterprise software vendors embracing things like MCP and A2A in their products.

[blog] A tour of Google Gemini libraries and tools. There’s not one tool or SDK to rule them all. That’s ok, as there are reasons to use each. I can imagine needing a Skill though, to help my AI agents make better choices on when to use each.

[blog] Stop Using /init for AGENTS.md. You might be getting whiplash from the different advice flying around. But we’re learning together live, so try and absorb the latest thinking.

[blog] Top Questions Answered: Getting Started with Antigravity. I couldn’t have answered all these, and it’s great to see what the community cares about.

[article] jQuery Releases v4: First Major Version in Almost 10 Years. A reminder that frameworks stick around, and even if you’ve moved on, millions of people still download this one every week!

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

Today’s list seemed to have a good number of items that could help you with planning work in your team: choosing frameworks, betting on databases, revisiting path to production, and skills training.

[article] What Developers Actually Need to Know Right Now. There’s a lot of wisdom in this post from Tim O’Reilly where he chatted with my colleague Addy.

[blog] How we rebuilt Next.js with AI in one week. This got me thinking about how often we’d now see people using AI to create compatibility layers or thinner versions of established frameworks.

[blog] I Taught My AI Coding Agent to Write YouTube Descriptions. AI-native teams and individuals ruthlessly attack toil with AI solutions. Karl has a creative and repeatable way now to make his video workflow simpler.

[article] How to adapt your skills for AI-driven development. Learn how to learn! There’s some useful advice here for taking a comprehensive approach to upskilling.

[blog] Google Renews Platinum Membership with the Linux Foundation, Continuing Its Ongoing Support for the Open Source Community. We’re not only a massive contributor to open source, but we also work hard to invest in the community at large.

[article] The End of CI/CD Pipelines: The Dawn of Agentic DevOps. I haven’t seen this point articulated like this, and I liked it. Either go all-in on agentic workflows, or stay with human-centered deterministic scripts. But don’t live in the squishy middle.

[blog] Firefly: Illuminating the path to nanosecond-level clock sync in the data center. It takes some crazy engineering to keep clocks in sync. This is a wild post about what we do.

[article] A Blog Post About COBOL Just Cost IBM $30 Billion. Here’s What Actually Happened. The services business will change. I don’t think we yet know how. But the markets has a quick reaction to Anthropic’s post about modernizing COBOL. Fintan has a good take too.

[blog] Flutter & Dart’s 2026 roadmap. Exciting plans for this pair. Generative UIs are going to be a big thing soon.

[blog] The Disintermediation of Databases. Rachel throws out a few hypotheses about the database market, and these feel spot on.

[blog] Powering the New Microsoft Agent Framework with Gemini 3.x. Use whatever framework gets you fired up. For .NET devs, that’ll likely be Microsoft’s Agent Framework. Good look here at making it work with non-OpenAI models.

[blog] Building a Faster GCP Kill Switch: Leveraging Cloud Monitoring Instead of Billing Data. In an ideal world, your cloud consumption would immediately turn into billing data. But the reality is trickier given how consumption data gets processed. So this person showed how to use monitoring metrics to get an early signal.

[article] The reason big tech is giving away AI agent frameworks. Sure, we’re all giving you frameworks so that it’s easier to use our respective stacks. Bet on higher order things like MCP and A2A to insulate yourself.

[blog] How to Manage Your Firestore Database with Natural Language via Firestore MCP Server: Step-by-Step Examples. I can’t say I’ve seen DBAs talking about doing database ops with agentic CLIs, but it seems like a solid use case. This isn’t necessarily ops, but gives a glimpse of what’s possible.

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

The first two items in this list are the two best things I’ve read in February. Just top tier, thought-provoking content that I can use right now. I bet you can too.

[blog] Building An Elite AI Engineering Culture In 2026. This should be a top 2 focus right now for most (every?) engineering team. Remake or eliminate handoffs, infuse quality into every step, and use AI tools for extreme leverage.

[article] Strategic choices: When both options are good. This could be the most practical and useful thing I read about “strategy” all year. Brilliant stuff from Jason.

[article] Skills are evolving too quickly for current training cycles, report says. Speed breaks all sorts of things. Including your training program. We need more skills training than ever, but it’s out of date every three months!

[blog] Great-looking UIs with Google Stitch and Google Antigravity. This is part of my go-to customer demo right now. I can build out an entire, attractive website in minutes using Antigravity and the Stitch MCP server.

[article] The Hidden Cost of Agentic Failure. More components (agents) means more failure points and hidden complexity. This post explores the tax and debt you incur with multiple agents.

[blog] Automating GCP Safely: How the Developer Knowledge MCP Server Powered Our Gemini CLI Skill. Fabulous post from Romin. He deftly tells a story that explains the problem (solving vague LLM requests) before going step by step through solutions (tools –> MCP –> Skills). Great read.

[blog] MLOps Pipeline with Vertex AI and Cloud Deploy on Google Cloud. I don’t even try to pretend I’m an ML engineer. But this is a very easy to follow post about how to set up a model training and serving pipeline. I’m tempted to try it myself now.

[blog] Decommission your legacy Apache Cassandra stack and build for the future with Spanner. A one-line code change gets you an entirely new database stack? Sounds like a good deal to me.

[blog] Simplify your AI workflow with autonomous embedding generation in BigQuery. Can’t beat that. Instead of a complex independent process for generating embeddings and updating rows, you can have BigQuery generate embeddings automatically for a table.

[blog] How Do I Enforce Quality Checks on AI-Generated Code in CI/CD? These look like the same things you’d check for any code! Just build good CI pipelines.

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

It’s good to be back at my home office today, even if I feel like I’m running at 70% after last week’s international adventure. Fun reading list today!

[article] Gemini CLI, Explained: Everything You Need to Know About Google’s Free AI Coding Agent. Excellent profile of Gemini CLI’s creator, and our momentum with the Gemini CLI.

[blog] Safeguarding Dynamic Configuration Changes at Scale. Configuration changes shouldn’t be exciting. Ideally, it’s boring and completely stress-free. That’s only the case if you’ve got a great system in place. Here’s what Airbnb does.

[blog] How I used Cursor to Migrate Frameworks. Good example of guiding a step-by-step upgrade of a web project’s dependencies.

[blog] Introducing the 185 Organizations for GSoC 2026. This amazing program brings new contributors to established open source projects for mentoring.

[blog] Better code, fewer tokens: Introducing Agent Skills for Firebase. Study best practices for your preferred product, tool, or framework. And then you can use something like Agent Skills to ensure you take advantage of those practices.

[article] Half the AI Agent Market Is One Category. The Rest Is Wide Open. There’s so much open territory for entrepreneur types. Tons of untapped domains that can take advantage of AI agents.

[blog] Teaching AI to read a map. Why can’t AI systems read maps very well? Their training sets don’t include much on the rules of navigation. Here’s new Google Research that shows a solution.

[article] State of Containers and Serverless. This latest report from Datadog has some new data points about who is doing what with Kubernetes, containers, and serverless runtimes.

[blog] Get ready for Google I/O 2026. Short post, but it’s about us getting the date out there for I/O.

[blog] The Agentic Web is Here: How WebMCP Transforms Websites into AI Toolkits. I’m getting more bullish on this browser feature every time I read something about it. Seems like a much more dependable way for AI agents to work with browser than Computer Use options.

[blog] This Week in Open Source #15. This is turning into quite the reliable look at what events and news are impacting the open source ecosystem.

[blog] Which web frameworks are most token-efficient for AI agents? Useful topic. When you’re building with your favorite web framework, does it take a lot of tokens to figure it out? Minimal frameworks performed well here.

[article] The Unreachable Engineering Managers. Slow is worse than off. An unreachable manager is much worse than one on vacation. We have to work hard to stay available and avoid becoming a bottleneck.

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

Finally flying home after a terrific week in India. Today I had the chance to talk to startup founders and community leaders to share Google’s plans and hear what they’re thinking about.

[blog] 5 key takeaways from the 2026 State of Software Delivery. Creating more code, and shipping less often? A small number of people are getting AI-generated code to prod quickly, but most are struggling to ship at speed.

[blog] Cloud Run Services: A Practical Guide to Getting More Bang for Your Buck. It’s one thing to use a service for a few minutes while experimenting. But, what types of configurations matter most when you’re sticking with a cloud service for a while? Good advice here.

[blog] 14 More Lessons from 14 years at Google. All lessons don’t apply to every person or situation, but I’m hard pressed to see how these aren’t relevant to all of us.

[article] How tech chiefs gauge ROI on AI. This looks like mature thinking about what tools can and cannot be responsible for.

[blog] The Secret Sauce of Reliable AI: Implementing Robust Pre/Post-Processing Hooks. Useful insight that I will try to remember more often. We have more capabilities at our disposal if we choose to use them!

[article] When Every Company Can Use the Same AI Models, Context Becomes a Competitive Advantage. Get your data management and context engineering efforts running at full speed if you hope to get legit value from AI.

[blog] What are the Trends in Agent Discoverability and Interoperability? This might seem like low-level content, but given how fast standards are changing, it’s important to know what’s going on.

[blog] How we reduced the size of our Agent Go binaries by up to 77%. It’s not hard for software to get bloated over time. Datadog has a nice post about how they made an intentional effort to strip out old dependencies, dead code, and other things that made their agent too hefty.

[article] Google Gemini 3.1 Pro first impressions: a ‘Deep Think Mini’ with adjustable reasoning on demand. Some thorough analysis of our latest model drop.

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

What a fun day talking to developers here in India. We had 1000+ doing hands-on workshops after a round of keynote talks. I might have learned as much as they did.

[blog] Gemini 3.1 Pro: A smarter model for your most complex tasks. A major leap on reasoning for complex problems! Pretty excellent benchmarks here.

[blog] Powering the next generation of agents with Google Cloud databases. This is the way. Don’t “hope” that the agent knows how to properly work with your systems. Use grounding. In this case, managed MCP servers for all our databases and documentation.

[blog] WebMCP – a much needed way to make agents play with rather than against the web. Don’t sleep on this. Every surface needs to be made agent-ready.

[article] AI Is Not a Library: Designing for Nondeterministic Dependencies. These are the types of things that human architects need to be aware of, and factoring into their decisions.

[article] How does development environment setup affect developer experience and productivity? This might be a secret productivity-killer among your technical staff. How much toil is there in setting up dev environments?

[blog] Your guide to Provisioned Throughput (PT) on Vertex AI. If you’re seriously using AI models, you want guarantees that the capacity is there when you need it. We’ve extended PT to more models, with great terms.

[article] A Guide to Which AI to Use in the Agentic Era. Life was easier in the old days of 2024. We just had chatbots. Now we’ve got agents (among other things) and the choice of model isn’t any easier.

[blog] Upskilling Antigravity for the Gemini API. This is a great justification for using Agent Skills. Make it easier to give models the correct context.

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

Today was a travel day as we bounced from Delhi to Bangalore. After a terrific customer visit, it’s nice to have a quiet evening before a 17 hour workday tomorrow.

[blog] New BigQuery global queries let you explore distributed data with a single SQL statement. The best managed services strip out complexity from your architecture. Here, we handle all the work of pulling distributed data together in one query.

[blog] Bias Toward Action. This doesn’t mean recklessly rushing forward. Have tight feedback loops and make small, responsible steps forward.

[article] Cloud and AWS cost consultant Duckbill expands to software, raises $7.75M for new Skyway platform. These folks have been around a while. Congrats on the raise and software ambitions.

[article] Multi-Language MCP Server Performance Benchmark. Do MCP servers perform best if written in Python, Go, Java, or JavaScript? The answer may surprise you.

[blog] Using go fix to modernize Go code. This is a fairly amazing capability just casually baked into the Go tooling.

[article] Anthropic releases Sonnet 4.6. Another attention-worthy release. As always, it’s available on your cloud of choice already.

[article] Most CFOs expect larger IT budgets, ‘collapsing’ staff growth: Gartner. I’m surprised. Tech spending up, but companies are getting more results from the same staffing levels. Or at least, hoping to.

[blog] Announcing America-India Connect and new investments to advance global AI access. The more connectivity the better for people of the world.

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

Short list today, as I was mostly occupied at our developer event in Delhi, and had little time to read. Tomorrow, on to Bangalore and more fun discussions.

[blog] How Generative and Agentic AI Shift Concern from Technical Debt to Cognitive Debt. Do devs need to slow down to avoid crushing cognitive debt that prevents their best work and harms the systems in the long term?

[article] Google Antigravity is the best fork of Microsoft VS Code and it’s not even close. I’m remarkably biased, but this doesn’t seem like a wacky opinion.

[blog] Run Voxtral Mini 4B Realtime on Google Cloud Run. Take Mistral’s small model and run in on a pay-as-you-go serverless stack.

[article] Open source maintainers are being targeted by AI agent as part of ‘reputation farming’. This headline would have made zero sense just twelve months ago.

[article] After all the hype, some AI experts don’t think OpenClaw is all that exciting. Maybe some of this autonomous agent stuff was oversold and isn’t (yet) as worthy of the given hype?

[blog] Gemini Interactions API: A Unified Interface for Models and Agents. This is a very powerful API and one that lets you offload a lot of machinery you had to maintain yourself.

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

I made it to Delhi without issue and took Sunday to adjust. Today, I was at our office in the area for an hackathon, and meetings with folks on my team. Tomorrow, we have our first big event of the week.

[article] It’s Time to Redesign How Product Teams Work. The types of changes called out here apply to many types of jobs, not just product team ones.

[article] Finding the Holy Grail of AI Agent UIs: From AI-Orchestrated Development to A2UI. Can we get away from writing so much frontend code? This article offers a good analysis of this problem space and solution.

[blog] Stop Closing the Door. Fix the House. Great stuff here from Angie. You can take your ball and go home, or you can make the game work in your favor. Open source maintainers should commit to the latter.

[blog] 3 Ways to Stop Wasting Your 1:1s With Your Manager. Great advice. Don’t just share status updates in your 1:1s. Bring professional development and a future focus to these chats.

[blog] Building an Autonomous Infra Agent using Vertex AI, ADK, and Google Managed remote MCP. Here’s a journey for building an agent that can stop, resize, and restart VMs on your behalf.

[blog] The Final Bottleneck. I’m glad that smart people are wrestling with this new AI reality. Nobody truly knows all the implications, and we need more people to think out loud.

[blog] State of the Designer 2026: Designers are leaning into the messy middle. How are designers feeling about the state we’re in? According to this latest Figma survey, pretty darn good.

[blog] Skills in ADK: On-Demand Instructions for your Agents. Excellent post that explains how to use Skills in your agent, and some good techniques for loading them correctly.

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