TL;DR

Calterra, a joint venture pairing Caledonian Climate’s peatland expertise with the Edinburgh AI firm New Gradient, has won £96,000 from Scottish Enterprise under its SMART Grant scheme. The 14-month project began this month and extends the platform from planning restoration to checking, years later, whether it worked. Calterra is also one of 12 firms in Cohort 10 of the AI Accelerator run by Edinburgh Innovations and the university.

The gap the money is aimed at

Restoration projects are easiest to assess at the beginning, when someone maps the land and decides what needs doing. Calterra has largely worked at that end so far, combining satellite observation with machine learning. Verifying recovery afterwards is harder, slower and more expensive — which is precisely why it tends not to happen often enough.

New Gradient’s founder and managing director Ewan McMillan put it plainly. The earlier work was mapping land and planning interventions, he said, and “this project is about what happens next”. The ambition is to follow habitat health “more frequently, more accurately and at a lower cost”, so that anyone paying for a restoration can see whether it delivered.

That last clause is the commercial point. Carbon credits issued against peatland restoration are only worth what the verification behind them is worth, and Caledonian Climate runs a substantial portfolio of Peatland Code projects.

Why the timing is deliberate

The new Climate Change Plan for Scotland sets a 2040 target of more than 400,000 restored hectares, and proposes a Peatland Standard covering consistency across restoration work, monitoring and verification included. A standard implies measurement at a scale field surveys cannot reach on their own.

The platform already has public backing behind it, including Innovate UK funding awarded jointly with Caledonian Climate and support from the UK Space Agency’s climate services programme.

Looking forward

This is a useful corrective to the shape of Scottish AI coverage this month, which has been dominated by calls to freeze data centre applications over energy and water demand. Here the technology is pointed at a climate target rather than competing with one. The grant is modest, and the question it has to answer is not whether the models work but whether verification gets cheap enough that it happens routinely — which is what the 2040 target quietly depends on.