Lightnews — Scholar-powered news

Karl Weinmeister @kweinmeister.bsky.social · 19h

What API call are you making 10x a day? 🤔 Turn it into a simple /command.

With Gemini CLI extensions, you can build your own shortcuts to speed up your work:
/fetch_jira ABC-123
/deploy_staging

Learn how easy it is to get started:
geminicli.com/docs/extensi...

3

Karl Weinmeister @kweinmeister.bsky.social · 1d

Has Gemini ever felt like it's losing focus as your conversation goes on? Naturally, more context means more topics to cover.

Use the /compress command to keep the Gemini CLI on track. It prunes the history without a full reboot.

Get started today with the Gemini CLI: npx @google/gemini-cli

1 1

Karl Weinmeister @kweinmeister.bsky.social · 5d

Not only is Nano Banana 🍌 production ready,
it now supports 10 aspect ratios!

Landscape: 21:9, 16:9, 4:3, 3:2
Square: 1:1
Portrait: 9:16, 3:4, 2:3
Flexible: 5:4, 4:5

Gemini 2.5 Flash Image now ready for production with new aspect ratios- Google Developers Blog

Our state-of-the-art image generation and editing model which has captured the imagination of the wo...

developers.googleblog.com

1

Karl Weinmeister @kweinmeister.bsky.social · 5d

Who’s your audience, and how are you offering value in your message?

AI tooling is immensely useful as a partner, but your engagement and insights remain essential.

hbr.org/2025/09/ai-g...

AI-Generated “Workslop” Is Destroying Productivity

Despite a surge in generative AI use across workplaces, most companies are seeing little measurable ROI. One possible reason is because AI tools are being used to produce “workslop”—content that appea...

hbr.org

Reposted by Karl Weinmeister

Remigiusz Samborski @rsamborski.bsky.social · 8d

Dublin 🇮🇪 we're coming!
Learn to build AI agents and deploy your MCP servers to production scale.

Register: goo.gle/accelerate-ai-dublin
Seats are limited!

Shir Meir Lador @kweinmeister.bsky.social @caseywest.bsky.social
#AI #AIagents #MCPServers #CloudRun #Workshop @GoogleCloudTech #DublinEvents

2 1

Karl Weinmeister @kweinmeister.bsky.social · 9d

Learn about Sparse Attention in DeepSeek-V3.2-Exp:
* O(L²) → O(L·k) with similar performance to V3.1 Terminus
* Lightning indexer scores previous tokens
* Top-k selector picks top 2k sparse tokens from 128k window

📄 Paper: github.com/deepseek-ai/...

🎬Video:
youtube.com/shorts/CLsju...

DeepSeek Sparse Attention Explained

YouTube video by Cloud with Karl

youtube.com

1 1

Karl Weinmeister @kweinmeister.bsky.social · 12d

You have four autonomy dials you can tune in your agentic AI system. Are you using them?

medium.com/google-cloud...

The Agency Spectrum: An AI Risk Management Framework

We can now build autonomous systems that pursue meaningful, high-level goals. Yet, for every inspiring success story, there is a…

medium.com

1

Karl Weinmeister @kweinmeister.bsky.social · 14d

All those pesky brackets making tokenization messy 😂

1

Karl Weinmeister @kweinmeister.bsky.social · 14d

When I have a choice, I’ve been picking it over JSON. The readability and comments are nice!

2

Karl Weinmeister @kweinmeister.bsky.social · 14d

Got it! Will pass on the feedback.

1 1

Karl Weinmeister @kweinmeister.bsky.social · 14d

@prietschka.bsky.social I appreciate your sense of humor 😀

I'm aware of this resolved issue about the diff view with light mode: github.com/google-gemin... Anything else specific about light mode support that stands out to you?

Regression due to #4739 Provide a better diff view on light-mode environments · Issue #5927 · google-gemini/gemini-cli

What happened? #4739 was fixed with #4747 and released in July. On light-mode environments the theme used is very much not readable: I am opening a new issue as I can not reopen the original one. W...

github.com

1 1

Karl Weinmeister @kweinmeister.bsky.social · 14d

If you're a Google AI Pro or Ultra subscriber, your daily limits for Gemini CLI and Code Assist just got a nice bump. Spend less time worrying about quotas, and more on building! 💻

blog.google/technology/d...

Google AI Pro and Ultra subscribers now get Gemini CLI and Gemini Code Assist with higher limits.

Google AI Pro and Ultra subscribers now get higher limits to Gemini CLI and Gemini Code Assist IDE extensions.

blog.google

1 4

Karl Weinmeister @kweinmeister.bsky.social · 14d

What really makes agentic software unique from traditional software? And is it all-or-nothing?

The essential concept is already there in the name: agency. You can control the degree of agency with tool design and human-in-the-loop patterns.

youtube.com/shorts/JIHfn...

Agentic AI: what makes it unique?

YouTube video by Cloud with Karl

youtube.com

Karl Weinmeister @kweinmeister.bsky.social · 19d

The Accelerate AI with Cloud Run tour is headed to Europe!
👉 Register at goo.gle/accelerate-ai

Join me and the Google team in a hands-on workshop near you:
📍 Dublin: Oct 29 goo.gle/accelerate-ai-dublin
📍 Munich: Oct 31 goo.gle/accelerate-ai-munich
📍 Paris: Nov 4 goo.gle/accelerate-ai-paris

2 2

Karl Weinmeister @kweinmeister.bsky.social · 22d

For a quick overview, check out the short video: www.youtube.com/shorts/1N5ub...

Ruff for Python Linting and Formatting

Speed up your Python development workflow with Ruff, the extremely fast Python linter and code formatter, written in Rust! In this video, I explore how Ruff can help you write cleaner, more…

www.youtube.com

Karl Weinmeister @kweinmeister.bsky.social · 22d

Yes, you still need code quality checks in the era of AI. Different prompts, context, and models lead to different outcomes.

Ruff unifies a litany of Python tools to do the job. Pro tip: send any issues it can't fix automatically to the Gemini CLI.

Read more: medium.com/google-cloud...

How to Write Better Python with Ruff on Google Cloud

Ruff can unify your code quality toolchain, accelerate CI/CD, and integrate with AI tooling for seamless Python development.

medium.com

1 1

Karl Weinmeister @kweinmeister.bsky.social · Sep 8

The blog post has all the code and step-by-step instructions: medium.com/@kweinmeiste...

Build with your Favorite Models from the Vertex AI Model Garden with LiteLLM

See Qwen 3 Coder in action as a Model-as-a-Service

medium.com

Karl Weinmeister @kweinmeister.bsky.social · Sep 8

Want to use models from Google's Vertex AI Model Garden with an OpenAI-compatible API?

My new video shows how to set up LiteLLM as a local proxy to do just that. Simplify your workflow and call models like Qwen, DeepSeek, and more through a unified interface.

www.youtube.com/shorts/ntI5A...

How to Connect to Vertex AI Models with a LiteLLM Proxy

Ready to explore the vast ecosystem of AI models in Google's Vertex AI Model Garden? This video shows you how to use LiteLLM as a unified bridge to access and experiment with a diverse range of…

www.youtube.com

1 1

Reposted by Karl Weinmeister

Guillaume Laforge @glaforge.dev · Sep 8

In-browser #AI semantic search 🧠 with Google's new #EmbeddingGemma embedding model and @hf.co's #Transformersjs

Gain enhanced privacy, zero server costs, low-latency results directly on your device.👇

glaforge.dev/posts/2025/0...

In-browser semantic search with EmbeddingGemma

A few days ago, Google DeepMind released a new embedding model based on the Gemma open weight model: EmbeddingGemma. With 308 million parameters, such a model is tiny enough to be able to run on edge ...

glaforge.dev

1 1 5

Karl Weinmeister @kweinmeister.bsky.social · Aug 28

Gemini CLI + Zed feels wonderfully natural. I got into the flow within minutes:

1. brew install --cask zed (on macOS)
2. Select "New Gemini CLI Thread"
3. Install Gemini CLI (if needed)

Learn more in this walkthrough: www.youtube.com/shorts/fAl--...

Agentic Development with Zed and Gemini CLI

Gemini CLI + Zed feels wonderfully natural. I got into the flow within minutes: 1. brew install --cask zed (on macOS) 2. Select "New Gemini CLI Thread" 3. Install Gemini CLI (if needed) Learn more…

www.youtube.com

1 3

Karl Weinmeister @kweinmeister.bsky.social · Aug 27

Do you need Docker installed to work with container images and repositories? Not necessarily.

With Skopeo, you can do LOTS of things without a daemon process or root access.

👉 Learn more: cloud.google.com/blog/topics/...

Five ways Skopeo can simplify your Google Cloud container workflow | Google Cloud Blog

Simplify Google Cloud container workflows with Skopeo. Learn 5 key ways this daemonless CLI tool enhances Artifact Registry management, image migration, CI/CD integration, and security for efficient,…

cloud.google.com

Karl Weinmeister @kweinmeister.bsky.social · Aug 26

The all-new Gemini 2.5 Flash Image model is built different. I had a blast mixing together multiple images and text!

👉 Try it out in AI studio: ai.dev
📖 Read all about it: developers.googleblog.com/en/introduci...

Karl Weinmeister @kweinmeister.bsky.social · Aug 25

"Relying on a single model forces an unnecessary compromise. Use a top-tier model for everything, and you pay a premium for power you don’t always need. Use a smaller model for everything, and you sacrifice quality on complex queries... Model routing is designed to solve this optimization problem."

A Developer’s Guide to Model Routing

Not long ago, building with LLMs meant picking one general-purpose model and sticking with it. Today, the landscape is flooded with…

medium.com

Karl Weinmeister @kweinmeister.bsky.social · Aug 19

With this getting started content, you'll learn how to:

✅ Scaffold a new Hono project using the official CLI.
✅ Build a complete CRUD API
✅ Connect to a serverless Firestore database
✅ Add production-grade middleware
✅ Implement type-safe input validation
✅ Deploy your application to Cloud Run

Karl Weinmeister @kweinmeister.bsky.social · Aug 19

💻 Grab the complete source code from GitHub: github.com/kweinmeister...

GitHub - kweinmeister/hono-cars-api: A RESTful API for managing car data, built with Hono, TypeScript, and Google Firestore.

A RESTful API for managing car data, built with Hono, TypeScript, and Google Firestore. - kweinmeister/hono-cars-api

github.com

1 1