FlieSense: the adhesive board, read by AI, not by eye.

In pest control, adhesive boards are an important source of field data: they show insect activity and help decide interventions. But the process is manual — the technician inspects the board, estimates how many insects were caught, records it and later turns it into a report. Slow, inconsistent and hard to scale across teams, clients and locations.

Agurys set out to change that: instead of estimating by eye, photograph the board and let AI do the reading — count, calculate occupancy, and hand the technician a structured result.

The challenge wasn't digitising a form. It was teaching software to read a technical object from the real world — an adhesive board full of insects, photographed in the field, with variable light and angles. Detecting, counting and supporting identification, repeatably, from a single photo. Without that, it was just another data-entry app.

We built the product around the technician's workflow.

The app (React Native) starts by scanning the board's QR code, validates the reference, guides image capture — single- or double-sided boards — and sends it to the AI engine. The technician reviews the result, completes the record (client, area, lamps, observations, date) and saves it.

Behind it, a computer vision engine analyses the photo: it detects the insects, counts them, supports identification and calculates the board's occupancy percentage. The heavy lifting of visual inspection no longer rides on the technician's eye.

And a backoffice (React, with a Strapi and PostgreSQL backend) gives the operation structure: management of technicians, clients, board references, insect types, counts and statistics, with PDF and CSV report export. It stops being just an app — it becomes a complete field tool.

Today FlieSense helps pest control professionals swap eyeballed estimates for AI-assisted reading: less repetitive work, more consistency across inspections, and field images that become structured information — reviewed, exportable and shareable.

This is the kind of product we like to build: AI applied to a specific operational problem — not as a generic layer, but as the engine that makes a technical process faster, clearer and easier to scale.