Environmental intelligence
Measurement beyond what a volunteer can carry.
Field evidence proves the work happened. Remote sensing shows what happened to the land afterwards — and the platform is careful to tell you which is which.
Data provenance
Know where every environmental metric came from.
Monitoring adapters can run against a real instrument or against a demonstration provider that produces plausible-looking numbers. Those numbers are useful for evaluating the product and dangerous in a sponsor's PDF, where a fabricated reading is indistinguishable from a measured one.
Reports carry only instrument-produced figures
Every observation records the provider that produced it. Anything published externally is checked against that record, and a section whose data cannot be attested is omitted from the report entirely rather than printed with a caveat nobody reads.
This is why a demonstration environment’s report may have no remote-sensing section. That is the guard working, not a missing feature.
Satellite
Vegetation monitoring over your mapped sites
EcoAttest integrates with Sentinel Hub for imagery over the boundaries you have already drawn, so monitoring needs no additional field work.
- Vegetation index (NDVI) readings over each site boundary
- Canopy coverage estimates from the same observations
- An observation history per site, not just a latest value
- Change detection across observations to show direction of travel
- Scheduled monitoring so the record builds without anyone remembering
Drone
Flight records and geo-tagged frames
Drone surveys are captured as flights against a site, with every frame stored, de-duplicated and positioned.
- Flight records against a specific site
- Geo-tagged frame ingest straight to secure storage
- Duplicate frames identified by content, not filename
- Coverage area and ground sample distance derived from the frames' own positions
Being straight with you: orthomosaic stitching is an integration point today, not a processing engine we run. Flight records, frame ingest, coverage and ground sample distance are real; a stitched orthomosaic requires connecting a processing provider. We would rather say so here than in your first month.
IoT
Continuous readings from the site itself
Where sensors are deployed, their readings become part of the same site record as the field evidence.
- A registry of sensors deployed at each site
- Device-authenticated telemetry ingestion
- Environmental readings stored as a time series
- Per-metric summaries alongside the raw record
AI
Turning field evidence into intelligence
AI here assists reviewers rather than replacing them. Every output is an estimate with a confidence score, and none of them approves anything.
Image quality assessment
Flags evidence that is too dark, blurred or otherwise unusable before a reviewer spends time on it.
AI-assisted tree counting
Produces an estimated count with a confidence score, offered to the reviewer as a starting point rather than as the answer.
Duplicate evidence detection
Identifies photographs already submitted elsewhere, including re-uploads of the same image under a different name.
Survival prediction
An explainable risk band per site from health, monitoring and vegetation signals — kept out of reports, which state measured figures only.
Anomaly indicators
Surfaces unusual submission patterns — bursts, concentration on one site, unusually high rejection rates — for a human to look at.
Human decision, always
No AI output approves a visit. Every one of these is an input to a reviewer who makes the call and is recorded as having made it.
Ask us what is real and what is an integration point.
We will go capability by capability, including the ones that need a vendor connected first.