# What is nexxus?

> nexxus connects the systems your company already runs, adds business modules on top and puts an AI assistant on every page. This page explains the idea and how the parts fit together.

nexxus is the AI-native business platform from Systaro. It reaches into the CRM, ERP, file servers and mailboxes a company already has, brings them together in one core and ships ready business modules on top. On every page sits the **AI Assistant**, which works on your real data with the rights of the person asking.

The short version from our website: *your people stop searching.*

## The problems it is built against

Most companies know these four. nexxus was designed around them from the first line of code.

<CardGrid>
  <Card title="Scattered data" icon="puzzle">
    Customers in the CRM, contracts on the file server, orders in the ERP. nexxus reaches into the existing systems through REST APIs or MCP instead of asking you to move everything.
  </Card>
  <Card title="Data protection" icon="warning">
    Sensitive data copied into external AI services. In nexxus you choose the providers, including European ones and models on your own servers, and tools ask for consent before they send data out.
  </Card>
  <Card title="Errors that look finished" icon="error">
    AI output that nobody checked. In nexxus the AI proposes and a person approves. It never writes to the database on its own.
  </Card>
  <Card title="Costs without an owner" icon="information">
    Many AI tools, nobody sees the bill. nexxus records every answer with its model, tokens and price, so you can see who spent what on which model.
  </Card>
</CardGrid>

## Four promises, one core

1. **AI-native, not bolted on.** The AI Assistant is there on every page and knows where you are. Open a customer and ask about their open invoices, and it already knows which customer you mean.
2. **Context is king.** The AI Assistant reads your company's knowledge, policies and processes, not just the prompt you typed.
3. **The AI proposes, a human approves.** Every change goes through a confirmation step that the server builds. The model never touches the database directly.
4. **Costs you can trace.** nexxus records the model, the tokens and the price of every answer. Admins see these figures under each answer and on the page *AI usage*, broken down by person, model and skill. Everyone else sees how expensive a model is before choosing it.

Here is what the third promise looks like when the AI Assistant prepares a change for you:

<Flow
  steps={[
    { icon: 'message-square-text', title: 'You ask', text: 'For example: draft a maintenance contract for this customer.' },
    { icon: 'sparkles', title: 'The AI Assistant prepares', text: 'It gathers the data it needs and proposes the change.', tone: 'lavender' },
    { icon: 'shield-check', title: 'The server checks', text: 'Input, current state and your rights are verified.', tone: 'cyan' },
    { icon: 'circle-check-big', title: 'You approve', text: 'Only after your click is the change carried out and logged.' },
  ]}
  caption="The AI Assistant can only propose what you are allowed to do yourself. The check runs on the server, not in the prompt."
/>

## How the parts fit together

nexxus runs in the browser and talks to one backend. The backend holds your data, the modules and the rights model. AI requests go through the **nexxus AI gateway**, which calls back into the backend for the tenant's model settings and tools. That is why every tool the AI Assistant uses passes the same permission check as the screen you click in.

<Hub
  inputsLabel="Connected"
  inputs={[
    { icon: 'building-2', title: 'Your systems', text: 'Microsoft 365, Google Workspace, Salesforce, HubSpot, Personio and more' },
    { icon: 'monitor', title: 'The nexxus app', text: 'In the browser, for your whole team' },
    { icon: 'plug-zap', title: 'Connectors', text: 'Actions in other tools, via REST or MCP' },
  ]}
  core={{
    title: 'nexxus',
    text: 'Backend with your data, the business modules and the rights model',
    items: ['Rights', 'Tenants', 'Audit log', 'Modules'],
  }}
  outputsLabel="AI"
  outputs={[
    { icon: 'route', title: 'AI gateway', text: 'Runs chats and rights-checked tools' },
    { icon: 'globe-lock', title: 'Providers you choose', text: 'IONOS, Scaleway, OpenAI, Anthropic, Google and more' },
    { icon: 'server', title: 'Your own models', text: 'Ollama, vLLM, LM Studio, any OpenAI-compatible endpoint' },
  ]}
  caption="Data is stored in PostgreSQL (with pgvector for search by meaning) and S3-compatible storage for files."
/>

## Security

These are the protections built into every nexxus installation:

- **The AI may only do what you may do.** Every tool the AI Assistant uses runs on the server with the rights of the person asking. Before it changes anything, the server checks the input, the state of the record and your permission, and you confirm.
- **Keys stay with the tenant.** API keys for AI providers are stored encrypted and never shown again after saving. One tenant's keys never reach another tenant.
- **You choose where the AI runs.** European providers such as IONOS in Germany and Scaleway in France, the large international providers, or your own models on your own servers through any OpenAI-compatible endpoint.
- **Consent before external tools.** A tool that sends data to an outside service asks before it does.
- **Tenants are kept apart.** Every record carries its tenant. nexxus takes the tenant from the signed-in person and filters every read by it.
- **Every change is logged.** The audit log records who changed which field, with the old and the new value.
- **Strong sign-in.** Two-factor authentication with an authenticator app, which an admin can make mandatory per role, and sign-in with Microsoft.
- **Hosting in Europe.** Data centres in Germany and Finland, or your own servers.

## Where it runs

It is the same platform wherever it runs. You choose between our cloud, a private instance or an installation on your own premises. Your data stays in European data centres or in your own building.

## What comes with it

nexxus ships ready modules for companies, contacts, projects and documents, plus the files area with the AI Assistant's document search. When a process is unique to your company, you add to it in two ways:

- **Apps** are small tools that live in your files. A business team builds them together with the AI Assistant, no code needed.
- **NXM modules** are full business modules with their own records, views and roles. Developers package them and install them in a tenant.

## Next steps

<CardGrid>
  <LinkCard title="Quickstart" description="Your first steps as an admin, from login to a first answer from the AI Assistant." href="/start/quickstart/" />
  <LinkCard title="Core concepts" description="Tenants, roles, modules, connectors and apps, and how they relate." href="/start/concepts/" />
</CardGrid>