d.velop AI | Knowledge Discovery

d.velop AI is the AI architecture of the intelligent d.velop content platform. d.velop pilot is the directly usable AI experience for knowledge retrieval: employees ask questions in natural language and receive precise answers from their own document inventory. Individual documents can be analysed and summarised; multiple pieces of content can be evaluated simultaneously. In addition, configurable assistants can be set up that access a defined subset of the company’s inventory — tailored to a department or topic.

d.velop AI works exclusively on the basis of your own content. Every answer is directly traceable to a source in your document inventory. This principle — known as Retrieval-Augmented Generation (RAG) — structurally prevents the AI from inventing statements or drawing on information outside your content. Employees can trace every answer back to its source and verify it independently.

The existing permission structures of your organisation apply directly to every AI process. What an employee is not permitted to see is also inaccessible to their AI assistant. Permissions are not configured separately for the AI; they are automatically inherited from d.velop documents. This applies to cross-departmental assistants and all queries across multiple documents.

d.velop pilot requires an existing d.velop documents repository and extends it directly with AI capabilities. Configuring an assistant requires no AI expertise: assign a name, set a search filter — done. In the background, a language model tailored to your content is created automatically. The assistant can then be embedded into any existing application landscape via a link.

Generic AI tools work on the basis of generally available training data and have no controlled access to your company’s content. d.velop AI accesses only your own content, respects your permission logic and anchors every statement in a verifiable source. Added to this is the embedding in the d.velop content platform: knowledge retrieval, processes and document management work from a single system, without interfaces to external services.

d.velop is a family-owned European company with no US parent company. What this means for data hosting depends on your choice. By default, d.velop provides suitable AI models for each use case — including providers outside Europe. Those who prefer a purely European setup will also find the right option with d.velop. User inputs and processed documents are transmitted in encrypted form. Which provider is actually used is documented in the data processing agreement. Sovereignty at d.velop means: you decide, we provide the infrastructure.

d.velop AI is designed for regulated environments. Permission structures, traceable source references and a complete audit trail meet the requirements of the EU AI Act, NIS2 and sector-specific requirements such as BaFin regulation in the financial sector or data protection requirements in healthcare. Organisations from the public sector, the insurance industry and manufacturing are already using d.velop AI in production.