Genkit

Announcing Genkit Dart (Preview) ✨ Build high-quality AI apps that run anywhere 📱💻 In this release: ✅ Support for Gemini, Claude, OpenAI ✅ Type-safe AI flows ✅ Dev UI for AI testing and traces ✅ Run code on client or server Read the blog → blog.dart.dev/announcing-gen…

@pavelgj Read the full writeup: dev.to/pavelgj/human-…

Human-in-the-loop is a security boundary. And security boundaries belong in code, not in prompts. 🔒 @pavelgj breaks down the two approval patterns for AI agents in Genkit—interrupts vs restartable tools—and when each is safe to use. 👇

@orimdominic_ It is available when using it with a remoteAgent!

@GenkitFramework I thought .chat() is deprecated

Conversational AI requires a lot of plumbing: message history, tool loops, streaming, and persistence. We're solving this with Genkit Agents! The new Agents API packages it all behind a single, consistent `chat()` interface. 👇

Read the full announcement on the Google Developers Blog to get started building agentic full-stack apps: developers.googleblog.com/build-agentic-…

With Genkit Agents, you get: ✨ Server- or client-managed state ✨ History branching from any snapshot ✨ Human-in-the-loop tool approvals ✨ Detached long-running tasks Available now in preview for Go and TypeScript. 👇

@golang @GenkitFramework @GoogleDevExpert 4/ All built on the public MLB Stats API. Not affiliated with @MLB, just a fan who wanted better stats. ⚾ Try it 👇 🌐 Web: diamondapp.xavidop.me 🖥️ CLI: brew install xavidop/tap/diamond ⭐ Code: github.com/xavidop/diamond @golang @genkitframework @googledevexpert

@leehack Check out Jhin’s announcement and the QuAI project: linkedin.com/posts/leehack_…

Hybrid AI in Flutter is now even easier! 📱✨ Community member @leehack just announced that QuAI is now feature-complete and truly free, powered by Llamadart, Genkit, and Gemma 4 for full on-device inference. Privacy-first, offline AI is here. 🚀 👇

Read the full announcement: genkit.dev/blog/dotprompt…

Genkit Dart 0.14.0 is here with Dotprompt support! 🎯 Manage your prompts in version-controlled `.prompt` files, decoupled from your app logic. Use Handlebars templates, multi-modal helpers, and tune everything in the Developer UI. 🚀 👇

*tap* *tap* Is this thing on? How many times do you find yourself waiting for some AI-based service to respond and wondering if it's stuck, thinking, or something else? Here's a @GenkitFramework pattern for streaming LLM thoughts back to the user: genkit.dev/blog/streaming…

Check out the original post and npm package details here: x.com/pavelgj/status…

Want to give your Genkit agents a real browser? 🌐 We built `genkitx-playwright`, a drop-in middleware that lets agents navigate, click, type, take screenshots & manage tabs. No more guessing CSS selectors—it uses accessibility snapshots! 👇️

Read the full blog post and get the open-source React components: genkit.dev/blog/streaming…

"Is it thinking or is it broken?" 🤔 Perceived latency is one of the biggest generative AI UX hurdles. We just published a new guide on building transparent AI chat UIs using Genkit. Learn how to stream the internal reasoning process (thoughts) of a model in real-time alongside the final response.

Check out the full release notes for details and getting started guides: github.com/genkit-ai/genk…

Genkit Python SDK v0.7.0 is officially live! 🐍🚀 We’re bringing pluggable middleware architecture to Python, giving you deep control over your generative AI workflows. ✨ Intercept the generation loop at key points with custom middleware 🛡️ Install pre-packaged, production ready middleware with `uv add genkit-plugin-middleware` Read more below👇️

Read the blog post: @denisov.shureg/hybrid-ai-in-flutter-routing-between-on-device-and-cloud-models-954da8f25373" target="_blank" rel="nofollow noopener">medium.com/@denisov.shure…

Hybrid AI is the future of mobile development. But how do you build it in Flutter without writing a mountain of duplicate code? 📱☁️ Google Developer Expert Sasha Denisov just published an excellent guide on routing between on-device and cloud models using Genkit Dart. The secret? Decoupling your routing policy from model runtimes. In Genkit, switching from cloud Gemini to local Gemma is as simple as swapping a single `model:` reference. Sasaha breaks down 4 routing strategies, from simple binary flags to local classifier models and custom fallback services. Read the full article here 👇