Overview

I designed Sepal, a mobile-first farm planning and team management app for cut flower producers

Cut flower farming is a small-team, high-complexity operation. What to plant, when to sow, who does what today, which bed needs attention, what grew there last year: most of it lives in one person's head. Sepal is an attempt to change that. Built with one engineer, from direct feedback from two active growers. Currently in prototype, test version shipping soon.

[Role]Sole product designer [Team]Antoine Richard (design), Daniel da Rocha (engineering) [Status]WIP prototype ready, test version shipping soon [Platform]Mobile (iOS-first)

Problem

Cut flower farming is a planning-heavy business. Most of it runs on memory and paper notes.

The information growers need to run a flower farm isn't complicated. Which bed has what in it. What stage each crop is at. Who is doing what today. What was planted there last year. But it lives in notebooks, WhatsApp threads, and whoever has been there longest. When someone is sick or leaves, a lot goes with them.

01

Planning is fragmented

Bed plans, task lists, and variety records live in different places and rarely connect

02

Planning stays personal

What to do on each bed, and why, lives in one person's head. Hard to share, hard to delegate.

03

Team coordination is fragmented

Daily task assignment happens through messages and word of mouth, with no record of what got done

04

Succession planning is hard

Timing sowings for continuous harvest means tracking multiple crops at different stages, in your head

05

Rotation warnings are missed

Without a record of what grew where, the same crop ends up in the same bed two seasons running

Why it matters

Local cut flower production is growing. The tools available to small farms haven't followed.

More growers are switching to local, seasonal production, supplying florists, markets, and direct customers. Most of these farms run on two or three people. The work is skilled and the margins are tight. A missed succession sowing or a rotation mistake doesn't just cost time. It costs a harvest.

+30%

growth in local cut flower production in France over the last five years

FranceAgriMer, 2024

80%

of small flower farms still plan primarily on paper or in generic tools like spreadsheets

User interviews, 2025

2–4

people on most small cut flower farms, making shared planning and team coordination essential

User interviews, 2025

Context & constraints

Growers work outside, in greenhouses, at the potting bench. Not at a desk.

That shaped everything about how the app was built. Fast to open, fast to record something, one thumb. But the same app also has to support the Sunday evening planning session: bed calendars, variety database, succession schedule for the next three months. Both modes are real. One can't be a worse version of the other.

Mobile-first, not mobile-adapted

The app was designed for a phone from the start, not a desktop interface squeezed down. Touch targets big enough to tap with dirty fingers. Input kept to a minimum.

Two modes, one tool

Daily task management and longer-term bed planning are both central to running a flower farm. The app had to support both without one feeling like an afterthought.

Small team, shared view

With two or three people on the farm, everyone needs to see the same thing. A plan that only one person can read isn't a shared plan.

Built for real users from day one

Users were involved from the first prototype. Every decision went through their actual workflow, not a hypothetical one.

Research

The first thing that became clear: growers don't think in task lists. They think in beds.

A grower doesn't open a to-do list in the morning. She thinks about Planche 3. What's in it, how far along it is, what needs to happen before the delivery on Thursday. Spatial and seasonal, not categorical. Every tool we looked at organised things by task type or date. That's the wrong starting point.

Insight 01

Growers think in beds and crops, not task types or dates. The bed is what they navigate from.

→ Beds are the primary navigation object. Tasks live inside beds, not the other way around.

Insight 02

The daily agenda is what growers actually use. What to do today, assigned to whom, in what order.

→ Today is the default screen. Not a calendar or a bed map.

Insight 03

A lot of what growers know isn't written down anywhere. Why a particular bed gets planted a certain way. What went wrong there two seasons ago. The app had to hold that knowledge, not replace it.

→ Bed notes, task logs, and variety records are built to capture that kind of knowledge. Not just structured fields.

Challenge

How do you build one tool that works in the field mid-task and at the kitchen table on a Sunday evening planning the season?

Without one mode feeling like a worse version of the other.

Process

Four decisions shaped the design.

Decision 01

Primary navigation

Should the app be organised around tasks, beds, or time?

→ Four tabs: Today, Beds, Library, Settings. Each does one thing. A bed opens its tasks. A variety shows which beds it is in. The links are there, but you don't have to follow them.

Decision 02

Task model

Should tasks be attached to beds, to people, or to dates?

→ All three. A task has a bed, an assignee, and a time. A grower who needs to desherber les dahlias needs to know which planche, who is going, and when.

Decision 03

Bed planning view

How should the bed's seasonal calendar be shown on a phone without losing the overview?

→ A horizontal scrollable timeline per bed, with crops shown as coloured segments. What is planted, when it was sown, what is coming next. Fits on a phone, still readable at a glance.

Decision 04

Rotation warnings

Should rotation warnings be automatic or left to the grower to check?

→ Automatic. The system knows what was planted where and when. Flagging a rotation risk costs nothing. Missing one costs a season.

Solution

A mobile-first farm tool with four clear modes

Today for the daily agenda. Beds for the farm layout and crop calendar. Library for varieties. Settings for the team. Each screen does one thing. The connections between them are there when you need them, invisible when you don't.

Screen 01

Today, the daily agenda at a glance

The first screen you see. Tasks grouped by time, each one showing the bed, the zone, and who is doing it. Mark done, defer, or open for detail. Tasks carried over from yesterday sit at the top. The weather is in the header because it affects what gets done: you don't spray biostimulant in full sun, and you don't harvest in the afternoon heat.

Screen 02

Beds, the whole farm on one screen

Each bed is a card: current crop, zone, status. Open it and you get the full picture: the timeline of what is planted, what stage it is at, the task log for this season, and the crop history going back as far as the grower has recorded. Rotation warnings appear on the card itself, not buried inside.

Screen 03

Library, the variety database

A searchable list of all varieties on the farm, pre-filled with agronomic basics: family, spacing, days to harvest, care notes. Growers add their own varieties and annotate them with what they have learned. From a variety card you can see which beds it is currently in and start planning a new sowing.

Pattern

Auto-generated tasks from the variety database

When a variety goes into a bed, the app generates the expected task sequence: sowing, transplanting, pinching, harvesting. Based on days-to-harvest and the current date. Growers edit or delete what doesn't fit. But not having to start from scratch every time is real time saved, especially on a farm running 50 beds and a dozen succession cycles.

Impact

The app is currently in prototype. Users have tested it across three rounds of feedback, which shaped the current version. We are now building the test app. Impact metrics will follow once real usage data is available.

Learnings

Working closely with active growers from the start made the feedback sharper than any formal research round. It also meant constantly checking whether a decision was right for this specific farm or for the wider use case.

The most useful part of the app turned out to be the notes. Not the planning features or the task system. Just writing down what happened on a bed so someone else can read it later. Most of what growers know about their farm is not written down anywhere. The app doesn't replace that knowledge. It gives it somewhere to live.

Mobile-first meant something specific here. Not just small screen. One hand, outdoors, mid-task. That constraint made every interaction simpler than it would have been if I had designed at a desk first.

More work