Start a project

Projects

Products we built and still run.

These are our own products, not client commissions. The team that built them is the team that builds yours.

Live market view ranking six retail chains by price, with the number of products tracked, active promotions and the average discount.

Consumer platform

Prices compared before anyone buys.

Shoppers search a product and see where it is cheapest, whether a promotion is real and which store wins on price overall.

450,000+products tracked

78stores followed

119,000promotions checked against the market

The problem

Shoppers cannot tell which store is really cheapest, and many promotions show a crossed-out price that was never charged.

How it works

  1. CollectProduct listings gathered from 78 online stores, supermarkets and pharmacies.
  2. MatchThe same product is recognized across stores, so prices are compared like for like.
  3. CheckEach promotion is compared with the real market price to flag fake discounts.
  4. PredictA model reads each product's price history to forecast the next 7, 15 and 30 days.

What we built

  • Product data collected across supermarkets, pharmacies and retailers
  • Price history and forecasts
  • Search across every store
  • Store rankings and promotion checks
  • An assistant that answers price questions in plain language
  • Public website and accounts

Built with Next.js, a price forecasting model trained on each product's history, and Llama 3.3 for the assistant.

Leaf photo where the vision model has framed four beetles, each labeled with its confidence score.

Computer vision

Crop problems found before they spread.

Drone cameras and vision models spot pests, water stress and early leaf disease before they can be seen from the edge of the field.

100+insect species identified

4Kimage resolution for analysis

The problem

By the time pests, disease or water stress can be seen from the edge of a field, part of the crop is already lost.

How it works

  1. PlanThe farmer chooses the fields, the dates and the analyses: disease, water stress or insects.
  2. FlyDrones capture high-resolution and multispectral images of each field.
  3. DetectVision models find insects, leaf disease and stress zones.
  4. ActMaps, indicators and alerts for each field, with treatment recommendations.

What we built

  • Detection models for insects and leaf disease
  • Multispectral image analysis
  • Field maps and early alerts
  • Flight planning for the drone service
  • Public website

Built with Next.js, multispectral drone imaging and detection models for insects and leaf disease.

Tell us what you need.
We will send a fixed quote.

Start a project