Genkit

It was awesome to get a chance to share what I'm hearing from talking to Genkit Go and Flutter Dev developers in this episode of the Agent Factory podcast! Thank you Luke Schlangen - it was a fun time! https://lnkd.in/gMQhEfHk

Genkit Dart hit 1.0 last week 🎉   I was at next.app devCon that week and updated the plugin right there, during the conference, to use it, take genkit_flutter_edge_ai 0.9.1 and genkit_hybrid 0.3.1 with genkit ^1.0.0 (a full report on this great conference is coming in a separate post) 😎 So you can now use Flutter Edge AI with the stable Genkit Dart:   • On-device: every model Flutter Edge AI supports works as an ordinary Genkit model, with streaming, tools and JSON output, plus embeddings, including EmbeddingGemma 2, fully offline. • Hybrid: genkit_hybrid puts an on-device model and a cloud model behind one routing policy, and your app still makes a single ai.generate call. A special thank-you to Christopher Gill from the Genkit team. At NextAppDevCon he walked through everything new in Genkit, and he mentioned Flutter Edge AI right from the stage. Thank you, Chris! 🙏   📦 https://lnkd.in/dmzM4KAV   📦 https://lnkd.in/dSKuC7KD   📖 flutteredge.ai/docs/genkit   #Flutter #Dart #Genkit #OnDeviceAI #EdgeAI #NextAppDevCon #FlutterGemma #FlutterEdgeAI

Want to run local AI models with Genkit Dart? Jhin Lee released genkit_llamadart 3.2.1, updated for Genkit 1.0! • Offline inference with GGUF & LiteRT-LM • Stronger tool & structured JSON support • Model inspection & live LoRA Check it out 👇 https://lnkd.in/gJ2GSafr

Announcing the stable 1.0 release of Genkit Dart, an open-source framework for building Gemini-powered features and agentic workflows with Dart and Flutter → https://goo.gle/474bCdr Since its preview launch earlier this year, feedback from the Dart and Flutter community has helped the Genkit team refine the APIs and expand the toolkit for production workloads.

Genkit Dart 1.0 is here! 🚀 Build production-ready, full-stack agentic apps with Dart and Flutter: • One API for Gemini, Claude, OpenAI & more • Strongly-typed structured outputs in Dart • Built-in tool calling, streaming & workflows Read the announcement blog 👇 https://lnkd.in/gnxYtTvT

genkit.dev has a fresh new look! 🚀 We rebuilt it around what Genkit does best and the real-world use cases we've heard our community has shipped with it. Explore it here 👇 https://genkit.dev

Genkit CLI & Dev UI 1.44.0 is out! 🚀 Key highlights: • A2UI Surface Rendering: Inline rendering and Surface Inspector in the Dev UI to inspect dynamic agent surfaces. • Mid-Turn Error Recovery: Resume or retry failed multi-step generation runs without losing intermediate tool calls. • Standardized kebab-case: Unified CLI command and flag syntax. • 60% Smaller Bundle: Dev UI footprint dropped from 9.2MB to 3.6MB for faster installs and loads. Full release notes: https://lnkd.in/gaXX6U4C

Genkit Python 0.12 is out now! 🚀 This release makes it easier to build interactive, multimodal AI apps in Python: • Generative UI with A2UI (genkit-a2ui): Let your agents respond with interactive forms, cards, and pickers instead of plain text. Pair a Genkit Python backend with JavaScript/TypeScript web apps or multi-platform Dart and Flutter apps using Genkit's client helpers. • Typed Pydantic streaming: generate_stream now yields partial Pydantic model instances on chunk.output as tokens arrive, giving you editor autocomplete and type-safe access for live UI updates. • Deep Research & Lyria 3 models: Run multi-step background research jobs with Google AI Deep Research and generate instrumental audio clips from text prompts with Lyria 3. • Easier mid-turn error recovery: Once a turn starts, ai.generate always returns a ModelResponse with error details and preserved message history so you can inspect failures or retry without losing progress. Install or upgrade: uv add genkit genkit-google-genai Read the full blog post and code examples: https://lnkd.in/gRZg6uzR

Which Gen AI framework should you choose for Go in 2026? 🚀 Xavier Portilla Edo published an in-depth, hands-on comparison of the top Go frameworks: Genkit Go, Eino, tRPC-Agent-Go, and more. Why Genkit Go stands out: • 3 levels of abstraction: low-level generation, typed flows, and agents in one SDK • Developer Experience: the Dev UI provides the fastest local development and trace loop in Go • Provider neutrality: switch between Gemini, Claude, and OpenAI in one line Read the full breakdown: https://lnkd.in/gMuwab3K

Testing human-in-the-loop agents? 🚀 When building agents that pause for human approval, feedback, or input (interrupts), debugging them shouldn't require deploying full applications. With the Genkit Developer UI, you can: • Isolate and step through human-in-the-loop workflows locally • Inspect agent state and pending interrupts in real time • Test resumption logic directly from the UI Check out Michael Doyle's demo walkthrough to see it in action: https://lnkd.in/gA_pNqBT