Introduction
The schnell.digital AI Kit is a platform for building automations — small programs that combine an AI model with your data, your applications, and your people. You do not write code: you assemble building blocks in a visual editor.
This page introduces the vocabulary used throughout the documentation. Two minutes here will save you a lot of clicking later.
The building blocks
Workspace
A workspace is a separate area for one team, customer, or department. It has its own users, its own automations, its own models, and its own quotas. Most organizations need only one workspace; agencies or consultancies often run several.
Automation
An automation is the thing you build. There are two flavors:
- A Workflow runs from start to finish on its own, like a small program. Example: every morning at 7:00, fetch yesterday's support tickets, summarize them, and email the summary to the team lead.
- An Agent has a conversation with a user. Example: an HR assistant a new colleague can chat with on day one.
Throughout the platform, both are called "automations". The difference matters only when you build them.
Trigger
A trigger decides when the automation runs: a fixed time, an incoming email, a webhook, a user starting a chat. Every automation has exactly one trigger.
Steps
After the trigger, the automation runs a sequence of steps. Each step does one thing — call an AI model, query a database, send an email, look something up in a knowledge base, write a file. Most automations have between two and ten steps.
Models
A model is an AI engine. The AI Kit does not run AI itself — it talks to providers like OpenAI, Anthropic, Mistral, or to a self-hosted Ollama. You connect at least one model to your workspace before you build automations.
Knowledge
A knowledge is a collection of your own documents, web pages, or database records that the AI can search. This is how the AI Kit answers questions like "What does our travel policy say about international flights?" without leaking your data to the public internet.
Integrations
Integrations are the building blocks for steps and triggers. They include things like "Send Email", "HTTP Request", "Query Database", "Cron Schedule", or "Watch Inbox". The catalog grows over time — see Automations → Integrations.
Inbox
The Inbox is where users actually use the AI Kit. They start agent conversations, pick up tasks that workflows hand to them, and read results. End users typically only see the Inbox.
Metrics
Every step that runs is measured — tokens, duration, cost, success or failure. The Metrics section lets you understand usage per automation.
Who does what
| Role | Typical tasks |
|---|---|
| End user | Uses the Inbox to chat with agents, picks up tasks from workflows. |
| Editor | Builds and edits automations, models, and knowledges. |
| Manager | Everything an Editor does, plus user-facing settings. |
| Admin | Everything: users, licensing, workspace settings, server health. |
| Monitor | Read-only view across the workspace — for auditors or stakeholders. |
The full role hierarchy is documented under Administration → Manage Users.
What to do next
- If you are setting up the platform for the first time, continue with Installation and Setup.
- If the platform is already running and you have just received an invitation, jump to Inbox → First Login.
- If you have logged in before and want to build your first automation, head to Automations → Create a New Automation.