Groundtruth maps how your business actually runs: who makes the calls, where the work lives, where it grinds, and how much change you can take on right now. You get that reading first. After that, any quote a vendor sends gets weighed against your business rather than against their own feature list.
No forms and no menus. You talk through your place the way you would tell a neighbour, and Groundtruth writes down what it hears as you go.
Worked example Mallee Grain is invented, so are John and his crew, and so are the three vendors further down. The conversation, the records it writes, the four readings and the capacity number are all running live in your browser, computed from the transcript beside them by the same open scoring engine that runs the method, version 1.0. None of it is a screenshot.
Tell me how the place runs. Start wherever you like.
It's a bit over 4,300 hectares out in the Wimmera. Wheat, barley, canola, a bit of chickpea and lentil. Eleven of us full time, more through harvest.
In the end the big calls, the marketing, the bank, the carbon stuff, they all come back to me.
So most things route through you. That can be a strength and a bottleneck at once.
Mark's our farm manager, been here longer than anyone. If he retires we lose half of what we know about these paddocks.
Michael's the one who got FieldView going, and the machine data goes straight in now. He'll chase down any new tool we throw at him.
Michael reads like a champion. I'll mark him as one.
And we bought the variable-rate controllers. They've been sitting half-calibrated for a year.
Gear that's paid for and idle tells me something about how much spare attention there is. Noted.
The spray diary still gets written up at the kitchen table at night. Same with the safety paperwork, nights and weekends in season.
And a couple of the blokes have asked me if this stuff means we won't need them.
Half the blokes have never really got their head around the precision gear. And between seeding and harvest, April through to December, no one's got time to learn anything new.
That worry is worth holding. Change that threatens people quietly stalls.
That's enough to see the shape of it. Here's your change capacity, and where I'd start.
| Type | Record | Status | Feeds | At |
|---|---|---|---|---|
| Person | John, owner | Owner | Alignment | 0:04 |
| Pattern | Everything routes through John | Watch | Alignment | 0:04 |
| Person | Mark, farm manager | Key | Roles | 0:09 |
| Friction | Knowledge walks with Mark | Risk | Roles | 0:09 |
| Champion | Michael | Strong | Readiness | 0:12 |
| System | Climate FieldView | Live | Readiness | 0:12 |
| History | Variable-rate gear idle | Watch | Readiness | 0:21 |
| Records | Spray diary on paper | Risk | Readiness | 0:26 |
| Friction | Paperwork eats nights | Watch | Roles | 0:26 |
| Concern | Are we still needed | Watch | Roles | 0:34 |
| Friction | Precision gear not understood | Watch | Readiness | 0:41 |
| Season | No room to learn, April to December | Watch | Readiness | 0:41 |
Nothing yet. The records arrive as John talks.
You talk through your place the way you would tell a neighbour. No forms. We listen, at your pace, in your own words.
Chapter 1As you talk, the picture builds: who decides, where the work lives, where it grinds, what people worry about. Every part of it is backed by something you actually said. It stays, so it becomes your memory of how the place really works.
Chapters 2 and 3It reads how much change you can take on right now. And when a vendor turns up, it weighs what they are selling against your business, not against their own features, and tells you plainly whether it fits, and what it would really cost to run.
Chapters 4 and 5Four plain questions, asked before a dollar is spent. Here they are read against John's conversation, with the line that decided each one.
Do the people who actually make the calls agree on where this is heading?
John and Michael want to push on, but every real call still routes through John, and the team has not agreed what it is all for beyond protecting margin. Placed in the forties: one keen party, and the deciders not heard agreeing.
"In the end the big calls all come back to me."
Has anyone worked out how the jobs change?
Who does what today is clear, but the paddock knowledge sits with Mark and no one has worked out how jobs change. Placed in the forties, moved down five because the crew has already said out loud they wonder if tech means fewer of them.
"If Mark walks out the door, a fair bit of what we know walks with him."
Can the place absorb this, the people and the systems both?
Good platforms are in place but half-used: the data flows in and no one looks, the variable-rate gear sits idle, and the spray diary is still done by hand at the kitchen table. Placed in the forties, moved up five on the installed base.
"They've been sitting half-calibrated for a year."
Does the plan fit this business, or was it copied from a template?
Not yet formed, and that is the honest answer after one conversation. The obvious first move sits where the carbon opportunity, the champion and a quiet season overlap.
"There might be thirty, forty grand a year in carbon if we could ever get the baseline sorted."
Four readings from one conversation. Nobody filled in a survey. Each score is the midpoint of a decade the evidence places it in, moved by five at most, once, with the reason shown. The rule is published.
Change has a budget. A business can only take on so much at once.
Groundtruth read John’s place at 44 out of 100. That number is what every quote gets weighed against, and here is what it is made of.
| What moved it | How much | What John said |
|---|
Every business starts at 60. Ten named things can move it, each at one of three fixed sizes, each with a line of evidence. Two scorers reading the same conversation land on the same number, or the method says why not. Then every quote that turns up is weighed against that number, not against the vendor's feature list. The handle below is there so you can see the rule work: it is not how the number is set.
Three invented vendors, each carrying a change load built from five things the quote asks of the business, at fixed sizes. Strong fit means the load sits at or under 60 per cent of capacity. Good fit, with conditions means it fits only if you stage it, or only its smallest version fits this year. Poor fit, for now means it asks for more change than you have this year, an overdraft, whatever the product's merits. The whole rule is published.
When a vendor turns up, Groundtruth weighs what they are selling against your business, not against their own features, and tells you plainly whether it fits, and what it would really cost to run.
Three invented quotes are loaded. Try each one against John's map.
Pick a quote Tap a vendor to see how it reads against John's map.
The four questions are not ours. Each carries decades of named research on why change succeeds or fails. Alignment traces to Cyert and March (1963) on who actually holds the decision, Cross and Parker (2004) on the informal network beating the org chart, and Kotter (1996) on the guiding coalition. Roles traces to Trist and Bamforth (1951), who found in the British coal mines that you redesign the work and the technology together or neither holds, and to Polanyi (1966) on the knowledge people hold but cannot easily tell. Readiness traces to Weiner (2009) and Armenakis, Harris and Mossholder (1993) on whether people are both willing and able, and to Cohen and Levinthal (1990) on absorbing only what you already have the footing for. Strategy traces to Goodhue and Thompson (1995): fit beats features. And the idea that change has a budget runs from Lewin (1947) through the change-saturation work to today.
As it happens, the research on how new things spread began on farms. In 1943 Ryan and Gross followed hybrid seed corn through two Iowa farming communities and found farmers adopted after a neighbour had proved it on nearby land, not when an expert told them to. Rogers built diffusion research on that study. We have added what is now understood about working alongside AI, held more carefully because it is newer: that people respond socially to machines that talk (Reeves and Nass, 1996), that trust has to be calibrated (Lee and See, 2004), and a field observation from one of the cohort behind this, a behavioural scientist who ran a 500-person user group and found people kept using the agent they liked over the one that was fastest or most accurate. That last one is unpublished and we treat it as a hypothesis to test, not a law. The method, and the evidence under it, are published in full, with the arithmetic as open-source code anyone can run.
For years the question was whether the tools were good enough. They are. So the thing that now decides whether AI pays off is not the model, it is the business: whether the people are ready, whether the plan fits, whether anyone has worked out what changes. Every tool before this one made you bend to the machine, learn its forms, its logins, its menus. This is the first one that can bend to you and simply listen. That is the opening. Whether you can take it depends on you, and that is what Groundtruth reads.
Two readers, one method. If you are a grower or a producer, it is a clear picture of your own place and an honest read on what is worth taking on, and what to leave for now. If you run a program or sit inside an RDC, the farm-level pictures roll up into a view across your whole network, so you can see where your people actually are, not where a survey says they should be.
The technology works now. That was never the hard part. Agriculture still has the lowest AI take-up of any Australian industry, at 19 per cent of organisations against 40 to 45 per cent elsewhere, and the usual explanations are that the tools are not ready or that growers are slow. Neither holds up on a farm. What you find there is gear that was paid for and sits idle, a platform used for a month of a two-year licence, and a steady line of vendors at the gate, none of them asking whether the place could absorb what they were selling. Nobody was measuring that. Groundtruth was built to measure it.
It starts from a conversation, not a survey. A person talks through their business the way they would tell a neighbour, and a map builds from what they say, every record carrying their own words. Four plain questions sit over the map, none of them new: each traces to decades of named research on why change lands or fails, and the founding study of how new ideas spread followed seed corn between Iowa farmers in 1943. Under the questions sits a budget. Change capacity is built from named factors at fixed sizes, with the evidence shown line by line, and every vendor quote is weighed against it as a load. The arithmetic is published as open-source code, so any reading can be checked by anyone with the same map. Nothing on the page, or in the method, shows a number without the working beside it.
The method is fully written and published, version 1.0: the conversation protocol, the scoring rules, the vendor drop, the costing convention, the data and consent rules, the calibration design and the research behind each question. It was built by a small cohort of industry thinkers across regional advisory, behavioural science, engineering and method, and it is available to any program, adviser or grower who wants to use it, teach it or build with it under an open licence. What it has not done yet is the field work: no calibration set has been scored by two independent people, so every reading is marked provisional until that is published. That cohort is the next step, and the form below is where it starts.
Groundtruth is a method, not a black box, and it is being built in the open. The thinking is written down and published. The worked example above runs on the published arithmetic. What it needs now is businesses to read and a second scorer to check the readings.