Discover the latest insights, guides, and updates on AI orchestration, real-time systems, and modern event-driven communication with DNotifier.

What LLM tool calling (function calling) is, what a tool call looks like, how dependent tool calls chain, and how to write tools a model uses correctly. With code.

Deep agents plan their work, hand pieces to subagents with fresh context and keep notes outside the prompt. How they work, and how to build one as a workflow.

A ReAct agent reasons, calls a tool, reads the result and repeats until it can answer. How the loop works, a 40-line example in JavaScript, and when to use it.

Where the agent, tools, secrets and UI should live when you add an AI agent to a real app. Patterns for auth, streaming, approval and limits, with code.

Build a working AI agent with tools, a knowledge base and a real UI. Step-by-step code for Node.js, React, Flutter and Dart, plus the mistakes to avoid.

Create a working AI agent that calls a real tool in about five minutes. One file of Node.js or Dart, explained line by line, with what to do next.

Part ten of the DNotifier × Ollama series — closing the full eight-provider arc: OpenAI through Ollama, one sendAI() pattern, and deliberate per-step routing.

An Ollama endpoint reachable from the internet is open inference for anyone. Lock it down with a reverse proxy, authentication, a VPN or a private network.

Part eight of the DNotifier × Ollama series — eight providers through one sendAI() pattern, capability versus ops tradeoffs, and mixed workflows.