How it works
Five steps between your question and your answer
Doc in the Bot is a retrieval pipeline with safety rails, not a chatbot improvising from memory.
Safety screening
Your wording is checked against deterministic emergency patterns before anything else happens. If it matches, an urgent-care prompt appears above the answer — it is never buried inside it.
Trusted source retrieval
We restrict the answer to an allowlisted registry of medical organizations: NIH, MedlinePlus, CDC, FDA, WHO, NICE, Cochrane, peer-reviewed literature and major academic centers. Blogs, forums, supplement marketing and content farms are never eligible.
Patient-language generation
The explanation is written for a patient, not a clinician, with a structure that always separates what is known, what may matter, and what to discuss with a physician.
Citation verification
Each citation must resolve to an allowlisted organization and a reachable URL. Anything that fails is removed, and the evidence rating drops accordingly. We would rather show fewer claims than an unverifiable one.
Evidence rating
Strong, Moderate, Limited or Unverified reflects the quality of the supporting evidence — never the likelihood that you have a condition.
What we will never do
- · Tell you that you have a condition.
- · Recommend a prescription medication or dose.
- · Let an advertiser influence a medical answer, its sources or its ranking.
- · Invent a citation, or keep one we could not verify.