This is a map of what the workshop group brainstormed, not a set of assignments. The useful move is to find the one line below that describes something you already do badly, then build the smallest version of it this week.
Ten kinds of work worth doing
Each card: the problem, a first version small enough to finish, and the kinds of tools people use once it grows up. You do not need any of those tools for the first version.
-
Maps you can actually use
Map fields, track rotations, monitor cover crops, look at imagery, plan grazing, and produce reports.
First version: one map of your fields with names attached, saved as a file you can open again next year.
-
Animal records that stay current
Monitor animal health and behaviour using sensors, cameras, or GPS. Track breeding, weight, movement, and pasture use.
First version: one page where you log an event per animal and see the last 90 days for that animal.
-
Watch the things you cannot watch
Track weather, soil moisture, water levels, and nutrient runoff. Get alerts and a clear view of what changed.
First version: one reading a day on one chart, with a threshold you choose and an alert when it is crossed.
-
Prices and margins in one place
Track prices, input costs, and market trends. Analyse profitability and compare scenarios for your operation or your members.
First version: a single sheet where you log one price a week and see the trend, with the source recorded next to it.
-
An assistant that knows your documents
Ask questions of your own data, regulations, research, or manuals, and build assistants for members, staff, or yourself.
First version: ask a question of a folder of your own documents and get an answer that shows which document it came from. The source matters more than the answer.
-
Somewhere members can find things
Portals for members, shared resources, event management, discussion, and connections across regions.
First version: one page of resources your members can reach without a login, replacing the email you send three times a year.
-
Reading submissions without drowning
Analyse submissions, summarize themes, map stakeholder networks, and prepare briefing material.
First version: summarize thirty submissions into themes, with a link back to the original text for every theme.
-
What exists within driving distance
Map local infrastructure such as processors, storage, veterinarians, and markets. Identify gaps and opportunities.
First version: a map of the six kinds of infrastructure you rely on, with the gaps visible at a glance.
-
The recurring task that does itself
Automate repetitive work, track maintenance, generate checklists, manage inventory, and send reminders.
First version: one checklist that generates itself on a schedule and gets sent to you, instead of being rebuilt each time.
-
Reporting that is not an annual panic
Track key indicators, visualize progress, and generate reports for programs, funders, or members.
First version: three indicators you already track, on one page, with one export you could send to a funder.
A simple way to begin
-
Name the real problem
Describe the friction in plain language before naming a technology. If the sentence contains a product name, try again.
-
Look for what already exists
Search open-source projects, existing services, research literature, and your own organization before building from zero. Most of what is on this page already exists in some form.
-
Define the smallest useful version
What could help one person or team with one task this week? That question produces better projects than "what could we build with AI".
-
Build in development
Use an AI to plan, prototype, and test. Keep changes small, documented, and under revision control. See Start here for the first session.
-
Verify before production
Check facts and sources, test the workflow, review permissions, and make sure you can recover if something breaks. See Checking the work.
Choosing between them
Pick the one where you already know what "better" looks like, and where being wrong is cheap. A tool for your own records is a better first project than anything that talks to members, sends email, or touches money. Not because those are impossible, but because they need the checking habits first.
If two ideas on this page appeal equally, start with the smaller one. You will learn more from something finished and slightly embarrassing than from something ambitious and half-built.