Case study · Farmland investment
Farmland LP
A digital twin of a $400M farmland fund manager, giving people and AI agents one trusted picture of the business to work from.
The situation
Farmland LP acquires farmland, converts it to organic and regenerative practices, and manages it for investors. It has $400 million under management across three funds, more than 15,000 acres, and a team of 60.
The firm is growing, and it wants to grow without burying that team in routine work. Its bet is that AI can take on the busywork one proven step at a time, so people stay focused on the decisions that matter.
That bet could not pay off on the data as it stood. Accounting, farm operations and investor information lived in separate systems and dozens of spreadsheets. The same farm, field or budget line was defined differently in each. An AI assistant pointed at that data gave confident, wrong answers, because nothing told it what the numbers meant.
The idea: a digital twin of the business
A digital twin is a maintained picture of the firm: its farms, its finances, and how they relate. People and AI agents read from the same picture, so they reach the same answer.
Two rules make it trustworthy.
- Every business term has one definition. “Acres” or “cost per acre” means the same thing in a report, a spreadsheet and a conversation with an AI assistant.
- Trying an idea is not the same as changing the record. A manager testing a smaller planting and a manager correcting the farm record may type the same number. The first is a proposal and stays with the plan. The second changes what the whole firm treats as fact, so it goes through a deliberate review.
The second rule matters most once AI is involved. An assistant asked to “change the acreage” needs to know which of those two things it is being asked to do, and whether it is allowed to.
What we built
- One governed platform. Operational, financial and investor data flow into a single data warehouse.
- A shared vocabulary. Each business term and rule is written down once, in a form both people and software can read.
- Applications for the daily work. Staff use purpose-built tools for budgeting, crop planning, harvest tracking and financial reporting. All of them read from the same definitions.
- A connection for AI assistants. Staff ask questions in plain English, and the assistant answers from the governed data, not from a guess.
- A team that knows how to use it. Through a few working sessions, staff learned to pair AI with the warehouse in place of building large one-off spreadsheets.
What changed
- A number can be followed back to its source: which land, which period, and whether it includes a planning assumption.
- Answers to plain-English questions tie to the accounting system.
- Reconciliation work that used to take weeks of manual effort is done in a fraction of the time.
What comes next
With the foundation in place, the firm is starting to hand routine steps to AI agents. Each workflow is mapped with the team that owns it. Each agent is supervised until its track record earns it more independence.
The takeaway
Farmland LP calls this its flywheel: connect data sources, model them once, serve them everywhere, document and reinvest. Each new project plugs into the same foundation and makes the next one easier. Nothing is a one-off.
Tools used