Pacific Northwest · Moscow, Idaho

Partners in the Northwest

We help county weed programs, conservation districts, parks, land trusts, and Tribal natural-resource departments find invasive plants earlier and send crews to the right ground. NatureNet.ai LLC runs the platform. You run the land.

Who we are

NatureNet.ai LLC is based in Moscow, Idaho. The platform and intellectual property sit with the Idaho LLC. Charitable and public projects are led by eligible applicants in the places where the work happens.

The problem we take on

Northwest weed programs and land managers cannot walk every acre they are responsible for. A drone flight they have already paid for can become a map of where to send the next crew — if someone turns the imagery into a review queue and a treatment-planning memo. That is the job.

Who we work with

Geography

Role Place
Company home Moscow, Idaho
Where we work Pacific Northwest — first ring Palouse, north Idaho, Lewiston, Pullman, Spokane. Oregon and western Washington when imagery is in hand.
Home-state footprint North Idaho flights (including Little Boulder Creek); University of Idaho research partnership
Published field evidence Minnesota buckthorn (Johnson Test), May 2026 — methods we still stand behind

Exact pilot sites are selected with each partner.

How partnerships work

Role Who
Grant applicant / fiscal lead Eligible 501(c)(3), government, Tribal nation, or fiscal sponsor
Technology partner NatureNet.ai LLC — platform, ML detection, maps, DataHub access, training
Field / ecology validation Independent forester, ecologist, or partner staff
Treatment / restoration Partner crews or contracted land managers

NatureNet.ai LLC owns the platform and IP. Charitable projects are separately governed. Services and software purchased from the LLC use arm’s-length, fair-market pricing. Conflict-of-interest disclosure and recusal apply when the same individuals sit on both sides.

What the program delivers

  1. Baseline aerial imagery, timed to the target’s best detection window (for buckthorn: late-fall leaf retention)
  2. NatureNet detection and a human review queue
  3. Human validation of a defensible sample — aerial candidates plus field transects for under-canopy stems
  4. Treatment- or management-ready maps
  5. Recording of management activity
  6. Re-survey at defined intervals — aerial change on visible stems plus repeat field transects
  7. Public report of methods, outcomes, and lessons

Data governance & public outputs

Why partnering compounds

The hardest open problem in aerial invasive detection is not the model architecture — it is that a detector trained on one flight can fail on the next one, because light, season, sensor, and site all shift underneath it. The fix is breadth: more validated flights across more conditions. That is a problem no single land manager can solve alone, and it is the reason we treat partnership as infrastructure rather than a sales channel.

Where partners agree to it, validated flights can contribute to a shared regional detector that every participating organization draws from — so a county's field verification improves the model a neighboring land trust runs next season. Participation is opt-in and defined in the subaward; a partner can use the platform without contributing anything. The public outputs from that work — protocols, uncertainty notes, training materials, aggregated benchmarks — are meant to outlast any individual project.

How the detector is deployed

NatureNet ships a human-in-the-loop detector for invasive woody plants, deployed as a review queue rather than an unsupervised auto-detector: the model proposes candidates, a person decides, and the two records are stored separately so a reviewer can always distinguish a prediction from a judgement. Every validation carries a name and a timestamp, which is what makes a recurrence figure defensible in a grant report.

Performance is assessed on your ground rather than ours. We bring field-validation results and methods directly into partner and funder conversations, and we will run the detector over a prospective partner's existing imagery so the assessment happens on a site they already know.

How aerial detection is scoped. Drones read the canopy surface, so for under-canopy targets like forest-interior buckthorn we aim flights at edges, gaps, and the late-fall leaf-retention window when buckthorn stays green after native leaf-drop, paired with field transects for the interior. Aerial survey prioritizes and extends field crews.

Read the program overview (PDF) · Research

How to partner

Write to hello@naturenet.ai (Moscow, ID) with the ground you steward, the target that worries you, and the timeline you are working against. One conversation is usually enough to scope what a first season would look like.