Blog Archive
Google ADK 2.0: Graph Workflows for Deterministic Agent Orchestration
Stop asking the LLM to do routing. ADK 2.0 graph workflows turn agent orchestration into deterministic code—cutting latency, token costs, and hallucinated execution paths.
The OpenAI Agents SDK: From Single Agent to Delegation at Scale
Learn the two core multi-agent patterns in the OpenAI Agents SDK—handoffs and agents as tools—and how to scale from a single agent to a delegating fleet of specialists.
Building Enterprise-Grade Multi-Agent Systems with the Microsoft Agent Framework
Take multi-agent systems from prototype to production with Microsoft Agent Framework 1.0. Learn graph-based orchestration patterns, checkpointing, human-in-the-loop, and deploying to Foundry Hosted Agents.
Function Calling and Structured Output for On-Device Agents with Gemma 4
Make on-device agents reliable with Gemma 4 function calling and constrained decoding. Learn structured output patterns, parse-error fallbacks, and how to build a trustworthy tool-calling loop with LiteRT-LM.
On-Device AI in the Browser: LiteRT-LM with WebGPU and WebAssembly
Run Gemma 4 LLM inference in the browser with LiteRT-LM, WebGPU, and WebAssembly. Learn the JavaScript API, serverless architecture, and how to build privacy-preserving web apps with zero backend.
Gemma 4 on Arm: SME2 and the Optimized On-Device AI Stack
How Arm SME2 instructions and KleidiAI accelerate Gemma 4 on-device AI. Learn about the hardware-software stack delivering 5.5x prefill speedups and enabling privacy-first edge experiences on Android.
On-Device Agentic Skills in Flutter with flutter_gemma_agent
Turn your Flutter app into an on-device AI agent with flutter_gemma_agent. Learn the SKILL.md catalog, four skill execution mechanisms, two-stage discovery, and the tool-calling loop that runs fully offline.
Mastering Edge AI on Raspberry Pi with LiteRT and Gemma
Deploy Gemma and LiteRT on Raspberry Pi 5 to build autonomous robots and local AI agents. Learn CPU and GPU inference, the Reachy Mini dual-processing pipeline, and the litert-lm CLI workflow.
Running Gemma 4 On-Device in a Multi-Agent Flutter App
Lessons learned running Gemma 4 E4B inside a Flutter app with multi-agent concurrency. Learn the platform channel bridge pattern, engine singleton design, serialization, and the multimodal constraints that bite in production.
Building a High-Performance On-Device LLM Client in Flutter with Qualcomm QNN
Learn how to build a production-quality on-device LLM client in Flutter that runs Gemma 4 on Qualcomm NPUs. Covers Dart FFI vs MediaPipe delegates, model delivery workflows, self-healing registries, and streaming reasoning UX.
Blazing Fast On-Device GenAI with LiteRT-LM
A deep dive into the LiteRT-LM runtime powering on-device GenAI. Learn how multi-token prediction, memory-mapped embeddings, session management, and constrained decoding deliver blazing-fast Gemma 4 inference on mobile.
Bringing Agentic Skills to the Edge with Gemma 4
Learn how to give on-device apps true agentic capabilities with Gemma 4, LiteRT-LM, and Agent Skills. Build multi-step autonomous workflows that plan, call tools, and act entirely on the device.