Permaculture tech
Permaculture AI, and what it can do for a site
How AI helps permaculture design: soil, water, climate and species data for a site in minutes, 3D maps of the plan, and why observation still comes first.
Permaculture starts with a year of observation. Watch where the water goes in a storm, where frost settles, which corner the wind hits, what grows by itself. No software replaces that, and any permaculture AI that tells you otherwise is selling something. But a lot of what a designer needs before the first walk is already sitting in public databases, scattered across a dozen agency websites in formats nobody enjoys. That part a machine can do well.
Green Fire builds tools that gather that data for a piece of land, put it on a map you can turn in your hand, and keep track of the design as it becomes real work. This page is an honest account of where AI helps a permaculture project, and where it should stay out of the way.
What a permaculture site assessment needs
A permaculture design rests on a few layers, usually in roughly this order of permanence: climate, landform, water, access, trees, structures, fencing, soil. Before a designer draws a single zone or sector, they want to know the rainfall and its season, the slope and aspect, where water enters and leaves the property, what the soil is, which plants and animals already live there, and what hazards the land faces, from flood to fire to drought.
For most sites in the United States, a good part of that has already been measured by somebody:
- Ecoregion and climate context. The EPA ecoregion maps describe the ecological region a site sits in, which tells you a great deal about what will thrive there.
- Watershed. The USGS Watershed Boundary Dataset shows which watershed and sub watershed the land drains into, which shapes every water decision.
- Stream flow. USGS river gauges show how nearby water behaves through the year.
- Soil. The USDA soil survey describes soil types, drainage and depth across most of the country.
- Species. Community science records from projects such as iNaturalist and GBIF show what has actually been seen growing and living nearby.
- Hazards and drought. Public hazard and drought data show the risks a design has to plan for.
Collecting all of that by hand takes a designer days. Our bioregional software, BioRegional OS, compiles it for a point on the map in a few minutes and keeps it offline, with the source and licence of every layer written down.
Where AI actually helps
Gathering and reading. An AI assistant can pull the regional data for a site, read the soil survey descriptions, and summarise what they mean in plain language: this soil drains slowly and sits wet in spring, that slope faces north, the creek below runs dry most Augusts. That is the dull half of a site analysis, and it is the half a machine does reliably.
Asking better questions. Given the data, an AI is good at listing what to go and check on the ground. Where does the water from that slope actually go? Is that soil map unit right, or is there a clay lens the survey missed? Good permaculture is a conversation with the land, and a well briefed list of questions makes the first walks count.
Showing the design. A permaculture plan on paper is hard for a family, a land trust or a donor to picture. We build interactive 3D maps of a site, with layers for water, buildings, gardens and access that can be switched on and off, so everyone can see the swales, the pond and the food forest before a shovel goes in. See the interactive 3D work.
Running the work. A design becomes hundreds of tasks across several seasons. In Civil X, our eco village software, every task sits on the 3D map of the land, with a before and after photo, so the community can watch the design turn into a landscape and nobody has to remember who planted what.
Carrying the admin. Grant applications, volunteer days, workshop sign ups and the emails that go with them. We build AI dashboards that draft that work and ask before sending anything. See AI for land projects.
Where AI should not lead
It does not replace observation. Public data is coarse. A soil survey is drawn at a scale where a whole field is one colour. A rainfall average hides the storm that matters. The AI can tell you what to look for. Only someone standing on the land can tell you what is there.
It does not decide the design. Permaculture design is a set of judgments about people as much as land: who lives there, how they move, what they love doing, what they can maintain. A model trained on text has read a lot about permaculture and has never stood in a field at dawn. Use it to draft options, never to choose.
It does not speak for the land's history. Many sites have Indigenous history, cultural places and relationships that do not appear in any database. Our software never downloads cultural information automatically and treats any map of territories as a prompt for a conversation with the people involved, never as permission.
It does not invent numbers. If a value was not measured, it should say so. A plant list or a yield figure that sounds confident and comes from nowhere is worse than no figure at all.
A sensible way to start
Pick the site. Compile the public data for it, so you walk the land already knowing the soil types, the watershed and what grows nearby. Spend the season observing, with an AI helping you log what you see and compare it with the data. Draw the design with people, then put it on a 3D map so everyone who has to approve, fund or build it can see it. Then break it into tasks and run the work where the whole community can see progress. That is permaculture with AI as a field assistant, not a designer.
It is also, to us, what solarpunk technology looks like in practice: small, local tools that help people understand and care for a real place.
Common questions
Can AI design a permaculture site?
AI can gather data, summarise it and draft design options, but it should not design a permaculture site on its own. Permaculture depends on long observation and on the needs of the people who live there. Use AI to prepare the site analysis and to communicate the plan, and keep the decisions with the designer and the community.
What data does a permaculture AI use?
A useful permaculture AI draws on climate and ecoregion maps, watershed boundaries, stream gauges, soil surveys, community science species records and hazard data. Green Fire's BioRegional OS compiles these public sources for a site and records where each one came from.
Is there free permaculture software?
BioRegional OS is free and open source and gives a site's ecoregion, watershed, water, soil and species data. It is not a drawing tool. Green Fire's 3D maps and Civil X task software for running a land project are paid services.
Can AI help with a food forest?
Yes, mostly with research and tracking. An AI can check which species have been recorded growing in your region, flag soil and water conditions that matter for tree choice, and keep a record of every planting and how it is doing. The species choice and layout should still come from observation and local knowledge.