Video courses

These tutorials show NappAI at work. The videos are in Spanish, and each one opens on YouTube in a new tab. Some were recorded with earlier versions of the app, so certain names and screens may differ from the current ones; for the details of each screen, see the app guide.

Guided videos from the documentation

Narrated videos recorded with the current application, each one a real case from start to finish. Each card opens the documentation page that explains it.

Getting started

Your first flow, step by step
A chat assistant from a blank project: components, connections, instructions, a test and a deployed version.

The AI Assistant builds a flow for you
You ask for a flow in your own words and it designs it, builds it, tests it and leaves it ready on the canvas.

Your storage
Upload files for your flows, see how they use and add to them, and download the results.

Team roles and permissions
What each role can do, choosing it when inviting, sharing a single flow and the activity log.

Agents and tools

Guardrails: rules the agent never breaks
An agent that cancelled orders on its own, with two rules: it refuses what is forbidden and never runs what breaks them. Voice agents too.

Approvals: the agent asks for permission
An agent that asks for approval before sending a quote, and how to answer: from the app, an email or another system.

Web Agent: an agent that works on the web
It reads a shop’s whole catalog as a table and logs in to a portal with a username and password, never seeing your passwords.

Python Code: every use
Packages, storage volumes, tools and models from your code, and your code as an agent’s tool.

Your own harness: a Deep Agent with tools, MCP and skills
An agent that prepares a business’s naming study: it searches the internet, checks domains with an MCP server and follows the agency’s method written as a skill.

MCP: use and publish tools
Give your agents MCP servers from the palette, and publish your flows as an MCP server for Claude, Cursor or VS Code.

A2A: agents working together
Publish an A2A agent and let another agent hand it tasks, in NappAI or on other platforms.

Voice agents

Voice agents: from idea to call
An agent that answers a clinic’s phone: configuration, a real call, publishing and the history with the recording.

RAG: answers from your documents and data

Advanced RAG: your documents and your data
A technical assistant that combines hybrid search with reranking, brand filters and an incident graph in one agent.

Hybrid search and reranking
Keywords and meaning at once, and Local Rerank so the exact part number comes first.

Metadata filters
Search only one brand’s documents, with a manual filter or one the agent fills in itself.

Self Query: the filter comes from the question
Tags with Metadata Tagger and a model that turns the question into a search filtered by brand and month.

Multi Query and compression
The question reworded several ways so the answer is not missed, and only the results that answer it.

Parent-child chunks
Search with small chunks and get the whole section back: a full procedure, step by step.

GraphRAG with Neo4j
Store records and documents in a graph, and an agent that queries it without writing Cypher: a technician’s incidents, similar cases and counts.

The RAG Agent
It answers citing its sources, follows up on earlier questions and says it does not know when it is not there.

RAG Evaluator: measure your RAG
It scores each answer against its sources and catches the one that makes up a fact, one by one or as a test set.

YouTube channel tutorials

1. First steps

• Discover NappAI
  A short introduction to the platform and what you can automate with it.

• The projects menu
  How to create, organise and open your flows from the projects screen.

• Add and connect components
  How to add components to the canvas and connect them.

• How to use templates
  How to start a project from a template.

2. Platform features

• Executions log
  How to review your flows’ runs and find where one failed.

• Version control
  How to save versions of a flow and work with them.

• How to use the Parse Data component
  How to format a component’s data with Parse Data.

• Component operations
  What a component’s operations are and how to choose the one it should perform.

• Unit testing flows
  How to test the parts of a flow inside the platform.

• Tool Mode
  How to use a component as an agent’s tool.

• Mapping Mode
  How to process a list of items with a component. In the current app this option is called Batch Mode.

• How to set up a Webhook
  How to receive in a flow the data another application sends.

• API integration
  How to run a flow from another application with the NappAI API.

• How to use the playground
  How to test a flow in the playground and check its result.

3. AI agents

• Why use an AI agent
  How an agent differs from an automation that always follows the same steps.

• Agent teams
  How to chain several agents to handle a process in several steps.

• Router agent
  How to classify a request and send it to whoever should handle it.

• How to use the Supervisor Agent
  How to coordinate a team of agents with the Supervisor Agent component.

• How to use Human in the Loop
  How to have a person approve certain actions before the agent carries them out.

• RAG agent
  How to make an agent answer from your own documents.

• Chunking
  How to split documents into fragments without losing context.

4. Use cases

• Managing a database
  A conversational agent that queries and manages a database.

• Gmail and Google Calendar
  An assistant that works with your email and your calendar. You can also follow it in writing in the email and calendar assistant use case.

• Data analysis
  A flow that analyses data and presents its conclusions.

• Build your own RAG
  A complete tutorial so an assistant answers from your documents.

• Automatic slide editing
  A flow that generates and edits slides.