Skip to content
Nappai Documentation

Chapter 1: Goal and components

In this use case you’ll build a personal assistant that answers questions about your email and your calendar. You’ll be able to write things like What meetings do I have on Monday 12 October? or Summarise the last emails I’ve received, and the assistant will check Gmail or Google Calendar to answer you.

How it works

The heart of the flow is an Agent: a component that reads your request, decides which tool it needs and writes the answer. For it to do that you’ll give it:

  • A language model, which is what understands and writes the text.
  • Two tools: one to read emails and another to check the calendar.

The finished flow: Chat Input connected to the agent, which uses Gmail and Google Calendar as tools and Gemini as its model, and its answer goes to Chat Output

The components

ComponentWhat you’ll use it for
Chat InputReceives what you type in the playground.
AgentUnderstands the request, uses the tools and writes the answer.
GeminiThe agent’s language model. You can use another provider, such as OpenAI or Anthropic, if you have its credential.
Gmail MessageThe tool that reads your emails.
Calendar EventThe tool that checks your calendar’s events.
Chat OutputShows the agent’s answer.

What you need

  • A Google account with Gmail and Google Calendar.
  • A Gemini API key, which you get in Google AI Studio.
  • A role that lets you create flows and credentials. See Roles and permissions.

In the next chapter you’ll create the credentials.