When a Japanese insurer wants to know how much biodiversity risk sits inside its corporate clients' supply chains, it now has a London startup on speed dial. In early 2026, MS&AD InterRisk Research & Consulting announced a tie-up with Gentian Limited to provide quantitative biodiversity impact and risk assessments for Japanese companies preparing nature-related disclosures [PR Times, 2026]. For a five-year-old company that started by mapping the moss on London rooftops, that is a meaningful jump in altitude.
Gentian, founded in 2020 and based in London, sells ecological assessments built from high-resolution satellite imagery and bespoke machine learning models [Nature Tech Collective]. The wedge product is called True Sight, a habitat mapping and site screening tool launched in August 2025 that the company pitches as a way to compress the cost and time of ecological site selection [Pro Landscaper UK, 2025]. The original use case was narrower: the technology was first designed to map vegetated roofs in dense urban areas, then widened to look at habitats at landscape scale in rural settings [NatWest Business]. That lineage matters, because rooftop sedum and a Hokkaido watershed look very different to a convolutional neural net, and the team has already done the unglamorous work of generalizing.
The bet
The thesis underneath Gentian is that biodiversity is following carbon's path from voluntary virtue to mandatory line item. The Taskforce on Nature-related Financial Disclosures (TNFD) framework is the lever. Once a Japanese megabank or a European food conglomerate has to attach a number to the ecological footprint of a soybean field or a solar farm site, somebody has to produce that number cheaply, repeatably, and at continental scale. Sending ecologists with clipboards does not cost out. Pointing a satellite at the parcel and running an AI model against ground-truthed training data does, at least in principle.
Gentian's product wedge, per its own description, is the fusion of high-resolution imagery with models tuned for ecological insight rather than generic land cover [Nature Tech Collective]. The MS&AD InterRisk partnership is the most concrete external validation of that pitch, because InterRisk is not a sustainability boutique. It is the risk research arm of one of Japan's largest insurance groups, and it is selling Gentian's outputs into the corporate disclosure workflow [PR Times, 2026].
Why it could be big
The accelerator graph is encouraging. Gentian has been associated with both Techstars and gener8tor [VCBacked, 2026], two programs that tend to push portfolio companies hard on enterprise sales discipline rather than research-grant comfort. Disclosed funding so far is modest: a $1.6 million pre-seed reported by AZOCleantech [AZOCleantech]. PitchBook and VCBacked profiles reference a larger $12.8 million figure dated December 10, 2024 [VCBacked, 2026], though the structure and lead of that round are not laid out in the captured sources.
Disclosed pre-seed | 1.6 | $M
Later round (reported) | 12.8 | $M
The market shape is the interesting part. Nature tech is roughly where carbon accounting sat in 2018: a handful of frameworks, a growing pile of regulation in Europe and Asia, and almost no incumbent doing the boring middleware of turning pixels into auditable numbers. If TNFD adoption follows even a fraction of the curve that TCFD did for climate, the addressable buyer set is every listed company with a meaningful land or supply chain footprint. That is a very large pool of CFOs who would rather pay a software vendor than staff an ecology team.
The competition
Gentian is not alone in the field. NatureMetrics works the eDNA angle, sampling water and soil for genetic traces of species presence. Dendra Systems comes from the restoration side. Pivotal and Natural Capital Research are also cited in the same competitive set [Business Insider]. Each has a different sensor stack and a different theory of where the defensible data layer sits. Gentian's bet is that satellite plus AI, with a habitat-typing model trained for ecological rather than agronomic outputs, is the cheapest per-hectare way to get to a number a corporate auditor will sign off on.
The honest counterfactual
What the bears will say is straightforward: biodiversity is harder to quantify from orbit than carbon is. A pixel can tell you canopy structure and approximate habitat class, but it cannot directly count beetles, and the gap between habitat proxy and species-level metric is exactly where regulatory standards are still being written. NatureMetrics' eDNA approach attacks that gap from the other side and may end up as the species-truth layer that every satellite vendor has to reconcile against [Business Insider]. The bull answer, visible in the InterRisk deal, is that corporate disclosure does not need beetle-level precision in year one. It needs a defensible, repeatable, auditable habitat baseline, and that is a problem that scales with imagery and compute, not fieldwork [PR Times, 2026].
Back of envelope
A traditional Phase 1 ecological survey in the UK runs roughly 1,500 to 4,000 pounds for a small site, call it 2,500 pounds for a 10-hectare parcel, or about 250 pounds per hectare (estimated, based on standard UK consultancy ranges). A satellite-plus-AI screening at scale should target something closer to 5 to 20 pounds per hectare to be interesting to a corporate buyer doing thousands of sites (estimated). Even at the high end of that range, the cost compression is roughly 12x. Multiply by, say, 50,000 hectares of screening across one large agribusiness portfolio and you are looking at the difference between a 12.5 million pound consulting bill and a 1 million pound software contract. That is the unit economics story Gentian needs to keep proving.
What to watch
The next twelve months are about whether the InterRisk partnership turns into a repeatable enterprise motion or stays a single marquee logo. Watch for a second insurance or banking partner outside Japan, ideally in the EU where the Corporate Sustainability Reporting Directive is forcing the same conversations. Watch for True Sight pricing to become public, because a published per-hectare or per-site rate would tell the market whether Gentian is selling software or selling reports. And watch for a Series A announcement that clarifies the December 2024 funding figure that has shown up in databases [VCBacked, 2026].
The company Gentian most has to beat is NatureMetrics. Both are racing to be the default data layer underneath corporate nature disclosure. NatureMetrics owns the species-level credibility. Gentian is betting that the buyer wants scale and cost first, and species detail second, and that the satellite is the only sensor that can deliver both.