Inspiration

The internet is shifting toward AI‑first discovery, and most websites aren’t built for machine interpretation. As a neurodivergent creator, clarity is essential to how I communicate — and I wanted a tool that exposes the gap between what humans website's content are saying and what AI systems actually extract. Semantica was born from that need: a simple, visual way to understand how content reads through an AI lens.

What it does

Semantica analyzes any webpage or app and returns: a semantic summary clarity and alignment scores extracted keywords machine‑vs‑human interpretation semantic gaps and improvement insights It’s powered by OpenRouter’s GPT‑4o‑mini with a stable fallback, and built entirely in MeDo’s visual workflow system.

How we built it

Designed the full interface in MeDo using the visual editor Integrated OpenRouter’s GPT‑4o‑mini for real‑time semantic analysis Added fallback JSON for reliability under API limits Connected UI components to workflow actions for instant results Optimized contrast and layout for ND‑friendly readability Tested everything directly inside MeDo without external hosting

Challenges we ran into

Getting meaningful, non‑generic output from a lightweight model Balancing real‑time API calls with a stable fallback Designing a readable, low‑friction interface under time pressure Switching from Groq to OpenRouter at the last minute Ensuring the analysis felt deep and structured despite model constraints

Accomplishments that we're proud of

Integrated a real LLM workflow inside MeDo Created a clean, ND‑friendly interface focused on clarity Delivered a stable demo with both real and fallback outputs Turned a long‑held idea into a working prototype Adapted quickly to last‑minute API changes without losing functionality

What we learned

MeDo made it possible to build Semantica fast. It handled: UI design workflow logic API integration testing visual refinement

What's next for Semantica

Stronger models for deeper semantic interpretation Website and document ingestion AI‑native rewrite mode ND‑friendly rewrite mode Agent‑readability scoring

Built With

• medo