If you want to run this yourself, here is the setup I would get in place before touching the code.

I am not going to reteach Microsoft Graph, Airtable, OpenAI, or Telegram from scratch in this case study. That would turn this into a bloated setup guide.

What I will do is tell you exactly what this project depends on, what shape those dependencies need to have, and what parts you should verify before you expect the service to run properly.

I am assuming you are technical enough to follow the repository, work through the official docs when needed, and connect the pieces without guessing what the system needs.

The five external pieces you need

Before this project can run, you need these five things in place:

1. a Microsoft account and a Microsoft Graph app registration
2. an Airtable base for state and job tracking
3. an Airtable MCP endpoint for job-record lookup
4. an OpenAI account with API access
5. a Telegram bot and a destination chat ID

That is the full external dependency surface of the project.

Use the repository together with this section

You should not treat the case study as the only setup source.

Use it together with the repository.

The repository already tells you a lot:

• .env.example shows the configuration surface
• the Airtable repositories show which tables are read and written
• the shared types show the data shape
• the README covers the local run flow