Subagents Highlight Durable Patterns in Agentic Engineering
I've enjoyed the ongoing look at "agentic engineering patterns" by @simonw. It can feel like nothing is settled in this space, but there's durability to many of these (https://t.co/TaHWvSyKbZ). His latest one is on subagents ... https://t.co/CpjxjDc1wN
Adding MCP Skills Cuts Agent Token Use 87%
My custom agent used 87% fewer tokens when I gave it Skills for its MCP tools https://t.co/Uki9krxFOY < I wanted to see how many tokens a custom AI agent would use in a variety of context settings. Surprising results? https://t.co/HphsVBWBEH
Code Writing Isn't Your Paid Core Responsibility
You're Not Paid to Write Code https://t.co/9HsV60TETy < good post from @milan_milanovic. Maybe you're getting tired of seeing posts on this topic, but remember that to change identities and habits, repetitive reminders and motivation are needed.
Remove PII to Safely Train High-Quality Models
"By helping to ensure that sensitive identifiers like personally identifiable information (PII) are selectively removed, you can feed your models high-quality data without the associated risks." https://t.co/6epQkSse1Q < these @googlecloud tools have gotten really good at this. https://t.co/H0ZyVDAzXO
AI Will Make SAP, ServiceNow, Salesforce More Programmable
"Ultimately, the destination with AI might not be to “replace SAP/ServiceNow/Salesforce,” but to make them more programmable and approachable." https://t.co/RkofGClRU2 < the world runs on SAP, and that may not change. But how we interact with it will. Good @a16z...
BigQuery Studio Gains Context‑Aware Editing and AI Discovery
We just turned on some new smarts in the @googlecloud BigQuery Studio interface. Now you get context-aware query editing (sees open query tabs), better resource discovery through natural language questions, and smarter troubleshooting. https://t.co/9ekJhzv0Ki https://t.co/SNntL6X6bB
Google Engineers Reveal AI's Real Impact on Delivery
Perk up when the DORA team does research into software delivery. The surveyed Google engineers to understand how AI tools impacted their work, and where they struggled. The advice is very actionable. https://t.co/PnaUMyiBLp https://t.co/mgU4WJb8SL
Skip Lengthy Guides—Use Agentic CLI for GKE Fixes
I don't want to read a giant troubleshooting guide, even a great one like this for GKE clusters. https://t.co/13E0PvzrMI Feed this as context into your agentic CLI (or use our @googlecloud Docs MCP) and send your agent down the right path faster.
Rethinking Engineering Team Size in the AI Era
What's the right size and shape of an engineering team when AI tools are part of the equation? Maybe an uncomfortable question, but we need to talk about it. @dblockdotorg has thoughts: https://t.co/4kw5EziS8r
Explore Pipe Syntax in BigQuery Sandbox for Free
Have you tried out pipe syntax instead of traditional SQL? I've only messed around with it a bit. I can see how it's an improvement for different types of queries. This post shows you how to try it out (at no...
Run AI Inference on Offline GKE Clusters
Sure you're running your Kubernetes cluster in a cloud for all sorts of reasons, but what if you want it without any internet access? Can it still do stuff? This post shows how you can still do AI inference on a...
AI Tracks Peer Help, Boosting Gratitude Bonuses at Google
Inside @google, we have a system for sending small bonuses to peers that helped us out. It's used often, and builds a culture of gratitude. We added an AI tool that scans your chats, emails, whatever and generates a report that...
AI Gives Cities 24‑Hour Flash Flood Warnings
Protecting cities with AI-driven flash flood forecasting https://t.co/CuGhkm1sPu < I really love that we do this sort of work. @GoogleResearch is helping us give up to 24 hours advance notice for flash flood risks. https://t.co/HIVZuV6ZWP
Forrester Launches AI Analysis Integrated Into Enterprise Tools
It's obvious that the big analyst firms need to make a transition. Faster analysis, more original data, better sense of what's happening on the ground. @forrester is now bringing their AI-based analysis service into existing enterprise tools. https://t.co/zklowwyBTs
Threat Actors Now Favor Third‑party Software Flaws over Credentials
"For the first time since we began publishing the CTHR in 2021, we observed a tactical pivot by threat actors. They’re now targeting third-party software vulnerabilities more than weak or missing credentials as the primary initial access vector." https://t.co/wSC5lPPGAZ
AI Adoption Stalls without Middle Manager Buy‑in
Not shockingly, execs still think everyone at their company is using AI all the time. The story on the ground is different. But notice the blurb about middle managers. When that tier isn't bought in, the initiative dies. https://t.co/Fys3E6zLm3
AI Boosts Productivity About 10%, Not Tenfold Overall
AI productivity gains are 10%, not 10x https://t.co/sjW5E3ob6P < We've said the same thing publicly. There are *tasks* that have 3x or 10x productivity gains, but it's not uniform across the whole day or entire value stream.
Master Agentic Engineering and Gemini CLI's Plan Mode
Seroter Daily Reading List – March 11, 2026 (#739): Today’s links look at the 8 levels of agentic engineering, using “plan mode” in the Gemini CLI, and how to quash your fear of messing up. https://t.co/3J5m1ku9c4
Gemini CLI's Plan Mode Safely Plans Changes Read‑Only
"With plan mode, @geminicli focuses first on analyzing your request, planning complex changes, understanding the code base or dependencies, and all this happens in a read-only mode safe from accidental changes or executions." https://t.co/TMVjSdxqOC https://t.co/HrGAbtrJ7J
AlphaGo Proves AI Can Decode Complex Physical Realities
"By proving it could navigate the massive search space of a Go board, AlphaGo demonstrated the potential for AI to help us better understand the vast complexities of the physical world." https://t.co/tDmqt6gWPQ < ten years of AlphaGo impact from @GoogleDeepMind
Secure Error Handling Prevents Sensitive Data Leaks
Do you think about security implications of your error handling in code? I do not, but that's because I'm a terrible developer who no longer ships production apps. But you probably are a good dev who avoids leaking info. Good @jetbrains...
Enterprise AI Cost Control Lies with Platform Engineering
"My original hybrid AI piece argued that developer workflows are the primary control plane for AI cost management. That’s true at the individual or small team level. But at enterprise scale, Layer 2C doesn’t belong to developers. It belongs to...
AI Assistants May Shift Developers Toward Preferred Languages
AI coding assistants may influence which languages developers use https://t.co/NUSmA6RF0d < seems possible, even likely. If you're polyglot, do you find that you favor one specific language for AI-generated code?
GoogleStartups Canada Accelerator Launches with 14 Startups
Our team drives @google's global accelerators, and we've just kicked off the seventh @GoogleStartups Accelerator: Canada cohort. We chosen fourteen outstanding Canadian-founded startups for this class. Learn more about each one: https://t.co/K5sbIaA7Mw
Preserve Data Relationships, Not Just Volume, for AI Success
"It is not just the width of your data that matters when it comes to getting great results from AI ... But capturing and preserving relationships is at the heart of all of it." https://t.co/XxQ9OTHEQ0 < @mipsytipsy with a great...
ADK Docs Now Include Built‑In Agent Skills
Hey, look at that. The popular Agent Development Kit just snuck some agent skills into their docs repo. These skills will help you create, evaluate, deploy, and observe ADK agents better. https://t.co/SblQXNAfAP https://t.co/huKLfdim8b
Gemini CLI Skill Detects and Refines AI Slop
Fixing AI Slop with a Skill in Gemini CLI https://t.co/xBTlmFrAFS < we can sniff out AI-generated text in many cases. @glaforge built a skill that uses known elements of "sloppiness" and rewrites text that sounds more like normal human writing.
Intensive AI Oversight Triggers Acute “Brain Fry” Fatigue
"The emergence of acute, overwhelming mental fatigue with intensive AI oversight—as distinct from burnout—adds new complexity to the picture." https://t.co/Dx7dlBRPZu < it's not burnout, but heavy AI users are experiencing "brain fry" as they exceed their cognitive limits
Google's gRPC Powers High‑Performance Services on GKE
Google still runs on gRPC, and many other companies embrace this high-performing RPC framework. Here's a good two-part series about using gRPC on @googlecloud Kubernetes Engine. https://t.co/Um5PI6PsQo https://t.co/HhT8uOKo1J https://t.co/GvyEdyt78G
Firebase Adds A/B Testing to Web Apps
Running experiments on your web app? Or wish you could? The powerful A/B Testing capability for @Firebase mobile devs is now available for web as well. https://t.co/vmNA9SlaFq https://t.co/z7xgjucwIJ
Scaling CI/CD Requires Far More Than Simple Pipelines
Me building a simple deployment pipeline for a silly app is very different than what it takes to manage CI/CD at scale. Here's a @semaphoreci post about how large companies do it ... https://t.co/rZx1GYWl3E https://t.co/tp8cbLsFZW
Sony Cuts Storage 91% and Costs Half with Spanner
"Sony Interactive recently rebuilt Entitlements from the ground up on Google Cloud Spanner, cutting storage by 91%, reducing costs by half (~48%), and completing the entire migration with zero downtime on a live production system." https://t.co/KpfYKfgaSE https://t.co/N2JO6jwxkn
Apply SRE Principles to Strengthen Security Practices
You can definitely apply SRE principles and practices to your security efforts. Here's a good post at things we do—eliminate toil, alert on symptoms, blameless postmortems, embrace gradual change—that you can do too. https://t.co/4lHNmUkQ52
Enterprise Zero‑Days Hit Record High, Near 50% Share
"Both the raw number (43) and proportion (48%) of vulnerabilities impacting enterprise technologies reached all-time highs, accounting for almost 50% of total zero-days exploited in 2025." https://t.co/EP6ycYhJJd < threat landscape is changing. You ready?
GKE Now Supports Custom-Metric Horizontal Autoscaling
Autoscaling is a magic part of cloud platforms, and a major reason picking use one. But scaling is often based on proxy metrics decided on by the vendor. We just lit up the ability for horizontal autoscaling on @googlecloud GKE based on...
Agent Development Kit Unlocks Powerful Cross‑tool Automations
Now that the Agent Development Kit connects to dozens of dev tooling services, you can do some pretty intriguing automations. Here's one perspective: https://t.co/YscWhGxpCK
New Google Workspace CLI Gains 9k Stars in Day
Sheesh, the repo is just a day+ old and you've already given us nearly 9,000 GitHub stars for the @GoogleWorkspace CLI. It's got 100+ agent skills, integrates with the Gemini CLI, and even bakes in an MCP server. https://t.co/6MQwOnrDeJ
Google’s AI‑Driven Playbook for Modern Security Threats
How Google approaches critical security topics, from fundamentals to AI https://t.co/4JCmvNxF8E < we're always learning things about new threats, new (and proven) techniques to respond with. @royalhansen takes a look here.
Timeless Software Engineering Practices Still Thrive Today
Coding has changed, no doubt, but software engineering itself is full of many durable ideas and practices. This post from @milan_milanovic shares a ton of lessons learned from the book "Software Engineering at Google." They still hold up! https://t.co/eOttYk6JAu
Set Safe Flag Defaults to Avoid Outage Risks
Safe flag defaults can prevent a simple mistake from turning into a major outage, says this @google Testing blog about setting safe defaults on your flag. Quick, useful advice ... https://t.co/SY9mNigoJm https://t.co/7tvZWW6Wql
Eventarc
Enterprise Service Buses offered great central governance but poor agility. Microservices had high agility but weak central governance. @googlecloud Eventarc Advanced aims for strong governance AND high agility. Central policy, distributed logic. https://t.co/dhKxsg21qK https://t.co/COgy20lps3
GVisor Isolates MCP Sessions, Preventing File Deletion Attacks
"How do you prevent the agent from accidentally (or maliciously) deleting files, uploading sensitive information to untrusted services, or consuming infinite compute?" @vladkol built an open source project that uses gVisor to isolate your MCP sessions. https://t.co/yMsHQbFWRb https://t.co/j5aEAxdrJi
Enterprise AI Adoption Lagging Behind Software Development Hype
Is everyone wrong about the timeline for AI changing software development? Depends on where you're looking. Enterprises don't move fast. Many are still getting going on "cloud migrations" and "DevOps." This might be different. Who knows. https://t.co/mNtDmqy7JW
Consultancies Remain Vital Despite AI Developer Tools
Yes, systems integrators and global consultancies need to make big changes. But a dev armed with Claude doesn't replace their value overnight. This post (and graphic) reminds us of how deep these vendors are with their clients ... https://t.co/KYjcwTYoxd https://t.co/RTDrMZUI3V
Goal‑Driven Agents Auto‑Orchestrate Tools Without Manual Steps
"Instead of manually specifying which model or tool to call and in what order, builders can now define a goal and let the agent determine the best path to reach it ..."
Install and Share Agent Skills Across Gemini CLI & Antigravity
How are you managing agent skills on your machine? And across tools? @kweinmeister shows us how to install and access skills in and across @geminicli and @antigravity ... https://t.co/RyIS9e6zMY https://t.co/N5gHc5GB5s
Enterprise MCP Adoption Surpasses Security Controls, Need Defense‑in‑depth
Shot: Enterprise MCP adoption is outpacing security controls https://t.co/B4FpJ7maqr Chaser: Securing AI Agents When Using Google Managed MCP Servers: A Defense-in-Depth Guide https://t.co/HBAXx8caUE
Spanner Adds Iceberg Lakehouse Support with 200× Faster Scans
"The columnar engine uses a specialized storage mechanism designed to accelerate analytical queries by speeding up scans up to 200 times on live operational data" The new @googlecloud Spanner capability means you can serve Iceberg lakehouse data ... https://t.co/dxmgEAI0cA https://t.co/TUe0vNnzfN
AI as Product Manager, Engineers Focus on Architecture
"Throughout this process, I'm partnering with AI to act as a product manager, but I'm also working as an engineer on the most important thing engineers can do: Technical architecture and how to break the problems down." https://t.co/KcFUc9RQXh < my...
Under 1% of Flaws Exploited, yet Weaponization Accelerates Dramatically
"Less than 1% of software vulnerabilities were exploited in the wild over the past year, but those flaws are being weaponized faster and on a larger scale than ever before" https://t.co/MJeKsYSj9R https://t.co/pvetZNrUmq