Beyond the smoke alarm: where AI is moving next in Australian forestry

When we last wrote about AI in Australian forestry, the focus was on cameras in the canopy. Pano AI deployments across plantation estates in Queensland, New South Wales, and South Australia. Triangulated detection of ignition points. Real reductions in time-to-suppression. That story is still progressing well. Forico in Tasmania, the New South Wales Forestry Corporation, and HQPlantations in Queensland all now run on AI-assisted detection networks, and the operational benefits are documented in plain language by the people running them.

Fire detection has become the public-facing AI story for Australian forestry because it is the one that has worked. It is visible. The outcome is unambiguous. A camera spots smoke at fifteen kilometres and a crew is moving inside the hour. It is easy to explain and easy to justify.

The other AI work in forestry is quieter. It is also where some of the more interesting questions for the sector now sit. Three areas are worth paying attention to over the next twelve months. Inventory and growth measurement. Supply-chain traceability. And mill-floor optimisation.

Inventory: what is in the stand, without sending someone in

Plantation inventory is one of those tasks that sounds simple and is not. To plan harvests, value a forest asset, or report against a sustainability standard, you need to know how many trees are in the stand, how big they are, what species they are, and how healthy they look. The traditional method is cruising. A team walks transects, measures sample plots, and extrapolates. It is accurate enough but it is slow, expensive, and limited to the parts of the estate you can practically walk.

Drone-based inventory has been promised for years. What has changed in 2026 is the AI layer that sits on top of the imagery. Australian plantation managers are starting to use systems that combine drone photogrammetry, LiDAR where it is available, and machine learning models trained to identify individual stems, estimate diameter, and flag stressed canopy. The output is not a perfect tally of every tree. It is a much better estimate than a cruise, generated over a much larger area, for a much smaller cost.

For an inland softwood operation in central Victoria or a hardwood plantation in the south west of Western Australia, the practical question is whether this is now worth doing in-house, contracting to a specialist, or waiting another year. The answer is usually contracting, for now. The drone hardware is not the hard part. The AI models are, and the better ones are sitting inside service-provider stacks rather than in any tool you can buy off a shelf.

Traceability: the European Union is forcing the issue

The European Union Deforestation Regulation comes into full effect for medium and small operators by the end of 2026. It applies to timber and timber products imported into the EU, and it requires due diligence statements that link each product to a specific plot of land and confirm that land has not been deforested since the end of 2020. Australia exports a meaningful amount of woodchip and pulp to Europe, and EUDR compliance is now part of what Australian exporters need to think about.

This is where AI is doing useful work that almost nobody outside the supply chain sees. Compliance platforms are combining satellite imagery, drone records, plantation management data, and shipment manifests into single due diligence documents. The AI part is the matching. Linking a container of woodchip leaving Albany or Burnie to a specific plantation, a specific harvest record, and a specific land-use history. Doing that manually for one shipment is possible. Doing it at the volume Australian exporters operate is not.

The cost of compliance has shifted onto the platforms that can ingest the data and produce the statement. For a regional plantation manager, the practical implication is to ask the questions early. Which platform is your customer using. What data are they asking for. What records do you need to be keeping now to make the compliance statement defensible in twelve months. The technology is not the obstacle. The records are.

The mill floor: vision systems that grade boards faster than people

Inside Australian sawmills, AI has been arriving on the mill floor through scanners and grading systems. Companies like Neural Grader and the SMARTI platform are deploying optical and structural scanning systems that grade boards in real time as they move down the line. The systems classify timber by visual defects, knot density, grain pattern, and structural integrity, and they do it faster and more consistently than a human grader at the end of a long shift.

The economic case is straightforward. Better grading lifts the recovered value of each log. A small percentage gain across a high-volume mill is meaningful, and the systems pay back on a timescale that mill operators find acceptable. The harder questions are about workforce. A vision system does not replace an experienced grader, but it changes what the grader does. The role shifts from inspection to exception handling and quality assurance. Where mills are training their staff to work alongside these systems rather than around them, the technology is sticking.

For smaller hardwood mills, the calculation is different. Capital cost is higher per unit of output, and the volume justifying the system is not always there. The technology will arrive in those operations eventually, but it will arrive later, and it will come through service models rather than outright purchase.

What this adds up to

Australian forestry's AI story is broader than fire detection, but the other parts are less visible because the outputs are less dramatic. Better inventory, defensible traceability, and improved mill yield are not the kind of headlines that travel. They are, however, where most of the operational value is going to be unlocked over the next two to three years.

For plantation managers, mill operators, and forestry contractors, the practical posture is the same one we have been recommending across other sectors. Be specific about the problem you are trying to solve. Understand what data you already have and what you would need to start collecting. Pick a small thing first. The interesting work in 2026 is not happening at the edge of what AI can theoretically do. It is happening in the gap between what the technology can already do and what is actually deployed in an Australian forest, a paddock, or a mill.

That gap is where the work is, and it is closing.