What this is, and firmly what it is not
This is a front-desk assistant. It answers the questions that flood a clinic's phone line: consultation timings, which department handles what, how to book or reschedule, what documents to bring, which insurers are accepted, where to park, whether reports are available online. It does not offer medical advice, interpret symptoms, or suggest treatment — and the design enforces that rather than merely requesting it in a prompt. Answers come only from your published content, and anything clinical, urgent or judgement-shaped escalates to your staff with the conversation attached.
Why grounding is non-negotiable in healthcare
A general-purpose chatbot bolted onto a clinic site will happily improvise about timings, fees or preparation instructions. In healthcare a confident wrong answer is not an inconvenience — it is a missed fast, a wasted trip, an unhappy patient at a reception desk. Flocci Chat retrieves from your pages first, shows which page an answer came from, and asks a clarifying question when a query is ambiguous ("fees?" — for which consultation, which doctor?) instead of averaging across your site.
Privacy posture
Only public pages are learned. Patient portals and logged-in areas carry data-passive-ingest="off", so nothing behind a login enters the knowledge base. Workspace data is isolated at the database level with PostgreSQL row-level security, and export and deletion tools are available for data-protection requests. Live screen state pushed by your own app is used for the current message only and never persisted.
Where it saves the most time
Repetition. The same twenty questions consume most of a reception phone line, mostly outside working hours. Answering those instantly, in the patient's own words, with the source page linked, gives your staff back the calls that genuinely need a person — and the unanswered-intent list tells you exactly which information your website should have published in the first place.