AURAMED (MEDI-BOT)

INSPIRATION:- We built AURAMED because asking basic health questions shouldn't feel clinical or intimidating. People often want quick, trustworthy guidance about symptoms, medications, or everyday precautions without wading through dense medical sites. AURAMED is inspired by that need: a warm, helpful digital companion that guides users toward clearer information and next steps while encouraging professional care when appropriate.

WHAT IT DOES;- AURAMED helps users in three main ways: triage (a friendly assessment of symptom urgency), medication lookup (clear info about common drugs and warnings), and precautions (practical, evidence-based tips to prevent or manage common conditions). You can type or upload a picture, get concise, plain-language answers, and see suggested next actions or sources. It’s designed to be an entry point fast, understandable, and safety-conscious.

HOW I BUILT IT:- I built AURAMED as a small web app using TypeScript and React, wired to a language model backend and a lightweight retrieval system for trusted sources. The UI is componentized so each flow (triage, pharmacy, precautions) is reusable and easy to update. The app uses structured prompts and JSON schemas to keep AI outputs predictable, and we added multilingual support so replies can be served in different languages based on user preference.

CHALLENGES I RAN INTO:- The biggest challenges were safety and reliability. Language models can sound confident even when uncertain, so we had to design prompts and schemas to limit hallucinations and force structured outputs. Handling images (medication photos) required careful file handling and clear messaging when identification wasn’t possible. We also balanced ease-of-use with the legal/ethical need to remind users this is not medical advice.

ACCOMPLISHMENT I'M PROUD OF THIS:-

A friendly, end-to-end UX that walks users from question to clear next steps.
Structured AI responses (JSON schemas) that make parsing and UI rendering robust.
Multilingual support so more people can get help in their native language.
Modular code (components + services) that makes iterating fast and safe.
Practical safety design: clear disclaimers, citation of sources where available, and explicit fallbacks when the model is uncertain.

WHAT I LEARNED I learned how important prompt design and schema validation are to trustworthy outputs. Small UX details — like showing a gentle disclaimer, making suggestions editable, and surfacing source snippets — greatly increase user trust. From an engineering perspective, building defensively (validate inputs, never expose secret keys to the browser, handle failures gracefully) made the app production-feasible faster than trying to patch issues later.

NEXT IMPLEMENTATION OF AURAMED (MEDI-BOT) Short-term:

Improve source citation and add automatic link previews for referenced guidance.
Add basic analytics (usage, edges where the model says “I don’t know”) to focus improvements.
Add automated tests for the critical flows (triage and medication parsing).

Medium-term:

Add user authentication and a simple history so people can revisit past interactions securely.
Integrate more trusted data sources and a review pipeline for the retrieval layer.
Expand accessibility and localization to cover more languages and reading levels.

Long-term:

Consider a clinician review mode or a lightweight escalation flow to connect users with professionals.
Certification or review for medical accuracy where appropriate.