MediBot

• 

  Medical

Inspiration

Our inspiration comes from the critical bottlenecks in hospital pharmacies and the total lack of accessible healthcare during late-night hours. Patients often endure long waiting lines just to receive basic medications, while pharmacy staff face immense daily workloads. We built MediBot as an intelligent, automated medication dispensing kiosk to directly solve this crisis. By seamlessly combining physical automated dispensing hardware with the advanced reasoning capabilities of OpenAI, we have created a 24/7, safe, and highly efficient solution that brings immediate medication access and automated health consultation straight into local communities.

What it does

MediBot is an intelligent, automated smart kiosk that digitizes medication dispensing and provides high-accuracy preliminary AI health screening. Users interact with the kiosk via an intuitive, natural language chat interface to describe their symptoms. The embedded AI acts as a smart clinical assistant, accurately recommending safe over-the-counter (OTC) medications or securely processing digital prescriptions. The system verifies the user's requirements, performs instant drug allergy cross-references, dispenses the exact medication physically from its secure internal inventory, and prints out simplified, personalized medication guidelines.

How we built it

Developed during the OpenAI Build Week, the production architecture of MediBot successfully bridges software intelligence with automated hardware execution:

Core AI Kiosk Interface: Powered by OpenAI APIs to handle the conversational UI, analyzing user symptoms and translating them into structured, error-free medication codes.

Knowledge Base & Safety (RAG): Deployed a robust Retrieval-Augmented Generation (RAG) pipeline integrated with a Vector Database containing verified pharmaceutical frameworks to eliminate AI hallucinations entirely.

Hardware Integration API: Developed a specialized API layer that directly connects the OpenAI reasoning engine with the kiosk's mechanical dispensing units (stepper motors, weight sensors, and live inventory tracking).

Guardrails & Authentication: Embedded rigid clinical safety layers that instantly flag critical medical red flags, halting medication dispensing for high-risk cases and immediately redirecting the user to emergency services.

Challenges we ran into

Our primary hurdle was ensuring absolute "Zero-Error Physical Execution." Because dealing with medicine leaves no room for mistakes, we spent extensive hours building rigorous prompt structures and parsing validations to guarantee the AI engine outputs strict, machine-readable JSON that perfectly aligns with the physical hardware actions. We also had to tackle the complexity of real-time drug allergy checks, which required designing a fail-safe backup verification database before any mechanical dispensing sequence could be triggered by the machine.

Accomplishments that we're proud of

Successfully engineered an operational system that bridges advanced generative AI with robust IoT/Kiosk hardware automation.

Created a highly reliable, clinical-grade triage trigger that safely restricts prescription-only or high-risk drugs while keeping common, essential treatments highly accessible.

Built a fully functioning end-to-end software prototype that processes a patient's conversational input and successfully queues the physical hardware for immediate dispensing.

What we learned

We learned that the future of practical health-tech is completely "Phygital"—where software intelligence perfectly drives physical machinery. Building this project taught us how to sanitize and transform loose AI natural language into strict, secure hardware commands. Furthermore, OpenAI’s latest reasoning models proved to be game-changing; their profound logic capabilities allowed us to move complex clinical validation rules directly into the context layer, significantly shrinking our backend code and speeding up deployment.

What's next for MediBot

MediBot is moving directly into its official rollout phase. Our immediate next steps involve deploying the physical kiosk units into select community pilots and medical centers for live clinical testing. We are currently finalizing our Multimodal integration, allowing the kiosk's hardware camera to securely scan physical doctor prescriptions and official IDs for age verification. Furthermore, we are actively expanding our API integrations with regional hospital networks to enable automatic synchronization with central patient electronic health records (EHR).

Built With

• api
• codex
• next.js