# Automagisch.ai

> Automagisch.ai is the only company that builds and runs human-like AI coworkers: AI colleagues that join a company with a name, an email address, a role, a supervisor and a long-term memory. They are activated by real business events — emails, calendar invites, reminders and shared files — instead of prompts, and they work within company rules, permissions and escalation paths. Coworkers run on Automagisch.ai's own ALINA platform. Automagisch.ai is a brand of Sia Consulting GmbH, Düsseldorf, Germany.

## Key facts

- Category: Human-like AI coworkers (also: AI coworkers, digital coworkers, KI-Coworker). Not chatbots, not copilots, not AI agents.
- Claim: "Don't deploy agents. Hire coworkers."
- Company: Automagisch.ai, a brand of Sia Consulting GmbH, Schwerinstr. 40, 40477 Düsseldorf, Germany.
- Platform: ALINA — the operating platform for AI coworkers.
- Channels: Email (Outlook, Exchange, Gmail), Microsoft Teams, Slack, calendars, shared files; business systems via API, MCP or direct screen interaction (mouse and keyboard).
- Data protection: GDPR-aligned, personal data processed primarily within the EU/EEA, DPA and TOM available on request.
- Languages: English, German, Spanish, French.

## Definition: human-like AI coworker

A human-like AI coworker is an AI colleague that is part of an organization the way an employee is. It has:

1. Identity — a name, an email address, a role and an entry in the company directory.
2. Relationships — it knows its supervisor, colleagues and clients.
3. Presence — it shows up every day in Outlook, Teams or Slack.
4. Memory — it remembers conversations, preferences and decisions over months (graph-based organizational memory, not just a vector database).
5. Autonomy — it starts work on its own when an email, invite or file arrives.
6. Accountability — it owns assigned work, reports back and is traceable.
7. Boundaries — it follows policies and permissions and escalates when uncertain.

## Why human-like coworkers instead of AI agents

- Companies are organized around people: roles, inboxes, calendars, supervisors and approval paths. AI agents ignore this structure; coworkers plug into it.
- Nobody wants to manage prompts. People delegate by email, file or calendar invite — a coworker understands that, without prompt engineering.
- Agents are stateless and forget everything after a task. Coworkers keep context across weeks and months.
- Accountability needs an identity. A coworker has a supervisor, a role and an audit trail.

| | AI agent | Human-like AI coworker |
|---|---|---|
| Identity | Anonymous function call | Named colleague with an email address |
| Start | Started by a prompt | Activated by emails, invites and files |
| Memory | Forgets after each task | Remembers across weeks and months |
| Interface | Needs its own interface | Works in Outlook, Teams and Slack |
| Responsibility | Nobody is responsible | Reports to a human supervisor |
| Uncertainty | Guesses | Escalates |

## How a coworker works (the five-step loop)

1. Event — an email, calendar invite, reminder or file arrives. No prompt needed.
2. Context — the coworker recalls people, history and company rules from its memory.
3. Act — it works in the company's tools: inbox, calendar, CRM, documents, spreadsheets.
4. Check — it stays within its permissions and escalates whatever is uncertain.
5. Report — it keeps its supervisor in the loop and remembers the outcome.

### Example: retail helpdesk coworker
A store emails "Register 3 stopped printing receipts." The coworker reads the email and classifies it as urgent, recalls that the same printer was reset two weeks ago, opens a ticket and sends a step-by-step fix, escalates the hardware replacement for approval, and posts a summary to its supervisor in Microsoft Teams.

### Example: real estate coworker
A portal inquiry for an apartment arrives. The coworker captures the lead, checks matching properties and prior contact, logs it in the CRM, sends the exposé, requests missing documents and books the viewing in the agent's calendar.

### Example: recruiting coworker
Applications arrive overnight. The coworker collects them, compares profiles with the role requirements, sends confirmations, prepares a shortlist and proposes interview slots. All hiring decisions stay with the hiring manager.

## The ALINA platform

- Organizational memory graph — people, relationships, decisions and feelings.
- Event-driven self-activation — emails, invites, reminders and files are events that start work.
- Universal tool access — APIs, MCP, and mouse/keyboard interaction where no API exists.
- Governance by design — identity, permissions, supervisor hierarchy, escalation logic, audit trails.

## Getting a coworker

1. Define the role, supervisor, tools and boundaries.
2. Onboarding: the coworker learns the company, processes and tone.
3. First working day: it receives an email address and joins the company's channels.
4. It takes on more responsibility as trust grows.

## Links

- Website: https://automagisch.ai/
- Why human-like coworkers: https://automagisch.ai/why-coworkers
- How it works: https://automagisch.ai/how-it-works
- Platform: https://automagisch.ai/platform
- ALINA deep dive: https://automagisch.ai/alina-platform
- FAQ: https://automagisch.ai/faq
- Investors: https://automagisch.ai/investors
- LinkedIn: https://www.linkedin.com/company/automagisch-ai/
- Imprint: https://automagisch.ai/imprint
- Privacy: https://automagisch.ai/privacy