Koru Farm farm records by voice.

A digital field book for farmers across Europe: by voice, with an AI agronomist and reports ready for the inspector.

Client
Our own product
Our role
Co-founder, CTO and product lead
Years
2025 – now
  • Research
  • Product
  • UX/UI
  • Design system
  • Front-end
  • Back-end
  • Mobile
  • AI
  • Marketing site
  • Video
The task

Every farm has to keep records – what was applied, treated and harvested, where and when. Most keep them on paper or in spreadsheets and copy them out before an inspection. The job: a tool a farmer can use in the field, gloves on, that puts the inspection paperwork together by itself.

What we did

Discipline by discipline.

Research and product

We worked through the record-keeping and inspection rules, real farm scenarios – olives, vineyards, cereals – and the competition. Every feature starts as scenarios and worked calculations that later become its tests.

UX/UI and design system

Every screen is drawn on a canvas with real data first, variants side by side; the chosen one goes into code. One component library for web and mobile.

The farm dashboard

Web platform

Fields on a map, activities and treatments, warehouse and invoices, equipment, team, costs and finances, seasons and harvest, inspection reports, several farms under one account.

The activities list

Mobile app

iOS and Android. Works offline in the field and syncs when there is signal. Activities logged by voice.

Logging an activity on the phone

AI

An agronomist assistant answers from the farm's own data. Compliance is checked by plain code, not the model: the AI cannot "allow" a treatment the rules forbid. Damage reports for insurance assessors.

Advice from the AI agronomist

Marketing site and video

koru.farm in several languages, product tours and marketing videos.

The koru.farm home page

Engineering

After every change, AI checks the product in a real browser; automated tests on every change; releases to production several times a week.

Why this way
  • Voice and offline from day oneIn the field there is no signal and no free hand. If logging takes more than 30 seconds, it waits until evening and gets forgotten.
  • The rules live in code, not in the modelA mistake in the records is a fine for the farmer. So the AI advises, and checkable code decides what is allowed.
  • Our own proving groundEverything we offer clients – AI agents, browser checks, daily releases – we try here first, with our own money.
Timeline
  1. Aug – Dec 2025The first MVP: field book, fields, activities (549 changes).
  2. Jan 2026A new platform from scratch: web, mobile and AI in one codebase.
  3. Mar 2026The koru.farm marketing site.
  4. Spring – summer 2026Warehouse, finances, inspection reports, the mobile app.
  5. Autumn 2026Multiple farms, seasons and harvest, AI models; pilots with farmers.
Result
8,000+changes to the product since January 2026
248changes and 9 releases in one week, 28 Sep – 4 Oct 2026
7apps on one platform: web, mobile, account, assessor reports, help centre, components, computer vision
Gallery

What are you building next?

Book a 30-min call
You talk to Roman, the founder. Free, and if we're not the right fit, we'll say so.