Cite, don't guess
Why
Trust is the product. A score nobody can trace is just another opinion.
How
Scores come only from public sources (no EWG or Mamavation data in any pipeline), and every flag points back to where it came from.
Nesvia
Reads every label in your home, scores each product against public science, and tells you what to swap first.
Can an app read the whole label of everything you own, score it honestly, and hand you one clear swap at a time?
People who want a less toxic home have apps that say “this is bad” and stop there. Nesvia does the next part: it audits the whole home, orders the swaps, and shows the score moving as you make them.
A product comes in as a shelf photo, a label photo, a barcode or a typed name. A vision model reads it, every ingredient is matched against public scientific and government sources, and the product gets a 0–100 score where higher means more concern. A product we cannot read stays unrated; it is never counted as clean.
How it works
Photo, barcode or name
Shelf photo, label photo, barcode, or a typed product name.
Vision model
Reads the product and its full ingredient list off the photo.
Ingredient matching
Each ingredient is resolved to a known substance.
Public sources
PubChem, EPA CompTox, openFDA recalls, Open Food Facts, Open Beauty Facts, USDA FoodData Central.
Product score 0–100
Higher means more concern. Unreadable stays unrated.
Home Score and swap plan
Highest concern and easiest swap first; the score falls as swaps land.
Decisions that shaped it
Unreadable is unrated
Why
Counting a product we could not read as clean would flatter the home and break trust.
How
The pipeline carries an explicit unrated state all the way to the Home Score instead of defaulting to zero.
One swap at a time
Why
People arrive overwhelmed. A list of forty problems makes them quit.
How
The plan orders products by concern and ease, so there is always exactly one next action, and the score visibly drops after it.
Built with
- Expo / React Native
- FastAPI
- Postgres
- Vision model
- PubChem
- EPA CompTox
- openFDA
- Open Food Facts
- Open Beauty Facts
- USDA FoodData Central