# AI Platform (BETA)

Automate internal processes with AI agents - set up, monitor and evaluate in one place.

One platform for every agent working in your organisation.

### Orchestration

Create agents, connect them to your systems and assign tasks - from a single check to a multi-step chain. Triggered on a schedule, by an event or by a request in the chat.

### Overview

Who is running, who is waiting, who has flagged something? The dashboard shows every run with its status, history and result. You can step in at any time.

### Analytics

Usage, run times, cost and error rates per agent - so you can see where automation delivers and where it needs adjusting.

### Documents & reports

An agent reads templates from SharePoint, fills them with data from your systems and files the finished report - the quarterly report, for instance, built the same way every time.

### Inbox & requests

Incoming mail is checked, categorised and linked to the matching case. Whatever needs a follow-up lands with the right person.

### Reconciliation between systems

Data from accounting, CRM and line-of-business applications is checked against each other on a regular basis. Discrepancies are reported before they become a problem.

An agent is configured, not programmed.

### Model & behaviour

You pick the language model and define how the agent responds - tone, format, context window. Open models such as Qwen run in Switzerland.

### Connections & permissions

Every connection - SharePoint or Abacus, say - is enabled individually, with read and write permissions per area. No more than the task requires.

### Rules & tests

Retries, time limits and stop conditions are settings, not lines of code. Test cases check before go-live that the agent does what it should.

All agents and their runs in one place.

### Runs & status

Every run shows up with its type, agent and status - resolved, stable, completed. Whatever is stuck, you see at once.

### Requests in the chat

Tasks can also be given directly: you write what needs doing and pick the model. The right agent takes over.

### Log & usage

Logs and reports show per agent what it did, how long it took and what it cost.

### Operated in Switzerland

Models, data and logs stay on Swiss infrastructure. Your data is not used to train models.

### Traceable

Every step an agent takes is logged. You see which data it read, which decision it made and what it triggered.

### Introduce step by step

You start with one process and one agent. What works gets extended - at your pace, with our support.

## Interested in the beta?

We keep adding organisations to the beta. Get in touch and we will work out together which process makes a good starting point.

[Request a meeting](/en/contact)