A behavioural health provider gave its staff an assistant that answers questions from the organisation's own documents

Clinical and admin staff needed answers from a large body of internal material: policies, intake procedures, programme details and service information. Finding anything meant knowing which document it lived in and searching through it, which favoured long-serving staff and slowed everyone else down. New starters asked colleagues instead, which pulled experienced people away from their own work. Out of hours there was often nobody to ask at all.

An assistant that reads the organisation's own documents before answering. Staff ask a question in normal language and get a direct answer along with a link to the source document, so they can check it themselves. Because it only answers from the organisation's real material, it cannot invent information, which matters a great deal in a clinical setting. New documents can be added at any time and become searchable immediately.

The technical name for this approach is retrieval augmented generation, which simply means the system looks up the relevant material first and answers from it, rather than relying on what a general AI model happens to know.

Built using n8n for the workflow, a searchable document store, and the Claude API for the answers.

Staff get answers in seconds at any hour, including nights and weekends. Experienced team members are interrupted far less. New starters get up to speed without needing to know which folder anything lives in.