🤖

No-Code AI Automation Course

Jul 31, 2025

Overview

This course guides beginners through building no-code AI agents and automations using Nadine (NADN), covering foundational concepts, practical builds, API integrations, prompting strategies, multi-agent architectures, and real-world deployment tips.

Course Introduction & Agenda

  • No coding experience is required; all concepts are explained step by step.
  • The course covers AI agents, workflow automation, APIs, tools, memory, prompting, webhooks, MCP servers, and deployment.
  • By the end, you will have built 15+ practical AI automations and workflows.

Foundations: AI Agents & Workflows

  • An AI agent is an autonomous system powered by a large language model (LLM) that can use tools to take action.
  • AI workflows are linear, predictable processes best for sequential tasks.
  • Use AI agents for unpredictable, decision-based tasks; use workflows for reliable, cost-effective, and scalable processes.
  • The core parts of an AI agent: input, LLM (brain), memory, instructions (system prompt), and tools.

Setting Up NADN & First Steps

  • Sign up for a free 14-day trial of NADN without requiring credit card info.
  • The workspace includes workflows, nodes, triggers, executions, and projects.
  • Triggers (manual, scheduled, webhook, etc.) start workflows; nodes handle actions, transformations, or AI calls.
  • Credentials (API keys) are needed for external integrations like OpenAI, Gmail, or Google Drive.
  • Data in NADN is visualized as schema, table, JSON, or binary (for files).

Data Types & Variables in NADN

  • Five main data types: string (text), number, boolean (true/false), array (list), object (dictionary/nested data).
  • JSON (JavaScript Object Notation) is the universal format for data and workflow templates in NADN.
  • Mapping and referencing variables between nodes is essential for workflow configuration.

Building Practical AI Workflows

  • Example workflows:
    • RAG (Retrieval Augmented Generation) chatbot using Pinecone (vector database), Google Drive, and OpenRouter.
    • Automated customer support via Gmail and a knowledge base.
    • LinkedIn content creation using Google Sheets, Tavi (web search), and AI agents.
  • Workflow building includes connecting triggers, integrating APIs, mapping data, and automating business tasks.

APIs & HTTP Requests

  • APIs let NADN connect to any online service (e.g., weather, Perplexity, image generation).
  • Use native integrations where possible; otherwise, use HTTP requests (GET/POST) with API keys, endpoints, headers, and body parameters.
  • API documentation and curl commands help configure HTTP requests in NADN.
  • Common API response codes: 200 (success), 400 (bad request), 401 (unauthorized), 500 (server error).

Advanced Tools: Webhooks, Images, Video, and Multi-Agent Systems

  • Webhooks allow NADN to listen for external events and respond asynchronously.
  • AI image generation (OpenAI, DALL-E) and video generation (Runway) can be integrated via APIs.
  • Human-in-the-loop nodes pause workflows for manual approval before taking action.
  • Multi-agent architectures (parent-child/orchestrator-subagent) enable complex delegations and reusable subworkflows.

Prompt Engineering for Agents

  • Effective prompting is crucial: start reactively with minimal prompts and add instructions based on errors.
  • System prompts should define role, tools, rules, and provide examples as necessary.
  • Output parsing ensures the agent returns structured, machine-readable data.
  • Prompt chaining, routing, parallelization, and evaluator-optimizer frameworks support specialized multi-agent workflows.

MCP Servers & Self-Hosting

  • MCP (Model Context Protocol) servers provide dynamic, flexible tool access for agents.
  • Deploying NADN via a cloud host (e.g., Alstio) allows installation of community MCP nodes and custom integrations.
  • Use self-hosted NADN for advanced scenarios involving MCP and community nodes.

Real-World Lessons & Deployment

  • Most AI agent demos are not production-ready; tailor solutions to your actual business processes.
  • Start with simple workflow automation; use agents only when dynamic decision-making is needed.
  • Always wireframe processes and plan data flow before building.
  • Combine no-code tools with custom code for scalability and flexibility.

Key Terms & Definitions

  • AI Agent — An autonomous system powered by a large language model, capable of decision-making and tool use.
  • Workflow — A linear, predefined sequence of automation steps.
  • Node — A functional building block in NADN (e.g., action, AI, data transformation).
  • Trigger — The event that starts a workflow (manual, timed, webhook, etc.).
  • Credential — API key or authorization required to connect to external apps.
  • JSON — A universal data format for structuring information and workflow templates.
  • Webhook — An endpoint that listens for and responds to external HTTP requests.
  • MCP Server — An external service that standardizes tool access and schema for AI agents.

Action Items / Next Steps

  • Complete all step-by-step workflow builds shared in the course.
  • Experiment with connecting new APIs using HTTP requests.
  • Practice prompt engineering: start with minimal prompts and add instructions based on observed behavior.
  • Set up your own self-hosted NADN instance if using MCP servers or community nodes.
  • Join the free school community for workflow templates and continued learning.
  • Optional: Join the paid community for live calls, classroom resources, and deeper support.

Full transcript