The AI assistant that knows your entire company

The AI assistant accesses all modules the logged-in user has access to: knowledge, CRM, finance, HR, ISMS, projects, marketing, assets and more. Ask 'Who manages the customer Müller GmbH and which contracts are currently running?' and the assistant combines CRM data, contract documents and accounting in a single answer. Permission awareness is built in: data you don't have access to, the assistant doesn't see either, neither as filter result nor as implicit background information. Multi-tenant isolated on Postgres with Row-Level Security. Tenant context is carried in the prompt header, no accidental cross-tenant leak. Context sources are listed transparently in the answer footer, you see where each piece of information comes from with a direct click to the source. For sensitive modules, additional extended audit logs are activated.

Upload PDFs, Word documents, Excel tables or Markdown files, the assistant extracts content and answers questions with concrete citation references. Example: 'What notice period does this supply contract have?' delivers the answer plus page number and verbatim quote. Works for multi-page contracts, technical specifications, audit reports or competitors' annual reports. Uploaded documents are stored multi-tenant isolated, processing runs EU-only via Mistral and AWS Bedrock in Frankfurt eu-central-1. OCR for scanned PDFs built in, table recognition with column mapping for Excel and PDF tables. Storage period per tenant configurable, default 90 days with automatic deletion after expiry, manual deletion possible at any time. Comparison function for multiple documents on the same query.

The assistant researches the web, e.g. competitive analyses, industry trends or regulatory updates. Sources are given with URL and date, you click directly to the original source and check yourself. Research queries run via an anonymised EU proxy in Frankfurt, your IP and tenant context are not passed to Google, Bing or other search engines. Results are stored in the chat history and are searchable later. Filter by source type (official authorities, trade press, Wikipedia, academic sources) configurable per query or as default per user. For regulatory research (e.g. BSI recommendations, GDPR rulings, NIS2 implementation law, IT Security Act) curated source lists are available, the assistant prioritises these over general web search and marks source reliability.

All chats are stored multi-tenant isolated, history is searchable by keyword, date or module reference. Mark important answers as favourites, organise chats in folders and tag them for later reuse. For team chats, colleagues see the history if permission exists, joint editing in real time possible with live cursor. Export per chat or history as PDF or Markdown possible, e.g. for documentation purposes or audit records. For regulatory audits (NIS2, ISO 27001) the export is audit-proof with hash chaining against subsequent manipulation. Chat storage period per tenant configurable, default 24 months with automatic deletion after expiry. For tax-relevant content the storage period is automatically extended to ten years under GoBD.

Standard model is Mistral Large 2, EU-hosted at Mistral itself in France. For more complex tasks the assistant uses Anthropic Claude via AWS Bedrock in Frankfurt eu-central-1, also EU-only. No OpenAI, no data outflow to the USA, no training on your data, contractually assured via DPA addenda with Mistral and AWS. Model selection per query automatically by cost-benefit profile, you don't have to take care of anything but can select manually per query. For especially sensitive tenants (e.g. KRITIS or defence-related companies) a 'Mistral-only' mode is also available, in which exclusively European models without US involvement are used, configurable per tenant. Model performance and cost per query are transparent in the dashboard, no hidden token consumption.

Save recurring prompts as templates, e.g. 'standard contract analysis', 'competitor research template' or 'BWA explanation for CEO'. Templates can be shared team-wide, everyone benefits from the power user's knowledge instead of having to come up with prompts themselves, learning effect in the team measurable over weeks. Templates with variables are possible, you fill placeholders per query, e.g. customer name, period or contract type, variables can also be mandatory fields with validation. Team management with permissions, versioning of templates, export and import as JSON for backup or migration between tenants for group structures. Rating function: team members give templates stars, best templates are highlighted in the dashboard with usage statistics. Prompt library with examples for common SMB use cases pre-installed, e.g. 'approve invoice', 'qualify supplier' or 'risk assessment for new software', you start with working templates and adjust them to your tenant specifics.