dAI Pro - 2026 Pricing, Features, Reviews & Alternatives

dAI Pro is an EU-hosted AI platform that answers visitor and employee questions using an organisation's own knowledge base, available as a TYPO3 extension (TYPO3 12.4, 13.4, and 14 on PHP 8.2+) or as a standalone web application. Content is ingested from up to 16 source types, including TYPO3 pages, PDFs, Office files, Confluence, SharePoint, Nextcloud, server folders, database queries, ticket systems, and mailboxes, and returned as grounded answers with source citations. A hybrid search engine combining pgvector vector search and full-text retrieval via Reciprocal Rank Fusion powers both conversational chat and website search from a single knowledge base. An automatic query classifier routes each question to retrieval-augmented generation (RAG), cache-augmented generation (CAG), or direct response, reducing chargeable model calls for recurring questions. The default language model is Mistral Large, hosted in the EU; per-collection configuration supports OpenAI, Anthropic, or Google Gemini with fallback models. Voice chat, permission-aware personal answers scoped to access groups, OCR for scanned PDFs, and human handover with full conversation history are included.

dAI Pro targets organisations with active data protection requirements: public authorities, mid-sized businesses, manufacturers, and any entity subject to works council or IT security oversight. The knowledge base is stored in the operator's own PostgreSQL database; only the question and relevant passages are transmitted to the language model, with no operator or visitor data used for model training. EU hosting eliminates US CLOUD Act exposure, and a data processing agreement is available from the demo stage. Every AI call is logged with token usage and cost in euros, exportable for audits and accounting. Real-time TYPO3 integration triggers immediate knowledge base updates on page saves, while a freshness monitor tracks all 16 source types and removes deleted content automatically. Multilingual deployments are supported through separate per-language knowledge bases with automatic language detection. An editorial copilot provides FAQ suggestions derived from real visitor questions, a gap radar, quality checks, and CMS backend text suggestions to reduce content maintenance effort.