Kanop
Kanop is a French geospatial intelligence platform that combines satellite imagery with deep-learning models to measure biomass, carbon stocks and land use change for nature-based carbon projects and for corporates reporting on their agricultural supply chains.
What is Kanop?
Kanop is a geospatial intelligence platform that measures carbon and biodiversity outcomes on land. It is published by RAIL SAS, a French company trading under the Kanop name, and it combines satellite imagery, deep-learning models and human expertise to serve two audiences: developers and financiers of nature-based carbon projects, and corporates that must account for the land-use footprint of their agricultural supply chains.
The measurement stack draws on optical imagery from Sentinel-2 (10-20 m, from 2015) and Landsat-8/9 (30 m, from 2013), radar from Sentinel-1 and PALSAR-2, and SRTM elevation data. A multi-encoder architecture fuses optical and radar signals through physics-informed prediction heads. The models were trained on 125 million hectares of LiDAR reference data spanning 76 countries, including 42 million hectares from 754 airborne campaigns and 83 million hectares from NASA's GEDI mission.
Outputs come in three layers: land use and land use change, carbon stocks and removals, and canopy and vegetation metrics at 10 to 30 metre resolution. Every estimate carries a confidence interval, 95% by default, propagated from pixel to polygon. Kanop publishes its accuracy openly: at site level the model reaches an RMSE of 41.7 t DM/ha and an R2 of 0.73 across 110 independent validation sites covering more than a million hectares; at 30 m pixel level the global weighted RMSE rises to 94.9 t DM/ha. Ground data is not required, though customer field plots can recalibrate outputs through Gaussian Process Regression.
External recognition is the platform's strongest credential. Verra has vetted Kanop as a Data Service Provider for VM0047 dynamic baselines, one of only three vetted for performance benchmarks, with more than 30 baselines delivered. Isometric selected it across both of its partner categories. Equitable Earth ranked it among top performers in an independent biomass benchmark.
Three access surfaces exist: a web dashboard, a REST API with public OpenAPI documentation, and an MCP server that pipes the data products into Claude, ChatGPT and Mistral's Le Chat. AI agents handle eligibility screening and report generation, usually within minutes to hours.
What it does
- Measure aboveground and belowground biomass and carbon stock change from satellite data, with a confidence interval on every estimate
- Build dynamic baselines that separate genuine carbon gains from pre-existing biomass
- Screen a project's eligibility against Verra, Gold Standard, Isometric and Equitable Earth methodologies
- Detect deforestation, forest degradation and land conversion across a sourcing network
- Quantify Scope 3 FLAG emissions and removals for GHG Protocol LSRS, SBTi FLAG, TNFD, CSRD and EUDR reporting
- Generate audit-ready reports and PDD greenhouse gas quantification sections
- Query the data products from Claude, ChatGPT or Mistral's Le Chat through the MCP server, or from your own stack through the API
When to use Kanop / When not to
A quick filter to help you decide if Kanop is the right fit.
When to use Kanop
- Carbon project developers running reforestation, REDD+, improved forest management or blue carbon projects who need methodology-aligned baselines and MRV deliverables
- Investors and offtakers screening or monitoring a portfolio of nature-based carbon assets who need independent verification of a developer's claims
- Corporate sustainability teams facing EUDR, GHG Protocol LSRS, SBTi FLAG, TNFD or CSRD deadlines on land-based emissions
- Procurement and sourcing teams tracking deforestation risk across cocoa, coffee, palm oil, soy, rubber or timber supplier networks
- Remote sensing analysts and GIS teams who want calibrated biomass and canopy layers delivered through an API rather than built in-house
When not to use Kanop
- Individuals and sole traders: the terms restrict access to legal entities and to professionals acting in a commercial capacity
- Buyers who need to compare prices before contacting a vendor, since no pricing is published anywhere and every deal goes through a quotation
- Teams needing plot-scale precision on very small parcels, where pixel-level error is roughly double the site-level figure
- Organisations requiring a signed DPA, a named subprocessor list or a recognised security certification before onboarding, none of which is published
- Anyone looking for a mobile app or a self-service sign-up, as the product is a web platform and an API reached through a sales conversation
How to use Kanop
A typical end-to-end flow, from setup to results.
- Book an introductory demo through the scheduling link, or use the dedicated form if you are a nature-based project developer
- Fill in the sign-up form on the application; an account opens automatically in your name
- Import your project boundaries as GeoJSON or shapefile into the web application, which is free during the trial period
- Let the AI agents guide methodology selection and run an eligibility screening against the relevant standard
- Review the dynamic baseline, historical trend analysis and ex-ante estimates produced for the area
- Receive a quotation built from the projects you imported, and accept it in writing within thirty days
- Create additional user accesses for your team once the quotation is accepted
- Track carbon stock change, land use change and canopy metrics from the dashboard as the project runs
- Generate an application token in the web app and call the REST API to pull results into your own systems
- Alternatively, connect the MCP server to Claude, ChatGPT or Le Chat and query the same data conversationally
Pros & Cons
Pros
- Vetted by Verra as a Data Service Provider for VM0047 dynamic baselines and one of only three vetted for performance benchmarks, with over 30 baselines delivered
- Accuracy is published with its validation protocol rather than asserted: RMSE, R2, sample size and a pixel-level figure visibly worse than the site-level one
- Every estimate ships with a confidence interval, and the models are designed to work without any field data, removing a major cost and delay
- API documentation is genuinely open: eight OpenAPI specifications are readable without authentication, which is rare in this market
- The MCP server lets analysts query forest carbon data from an AI assistant they already use, with no platform to log into
- Data portability is unrestricted: export as CSV, raster or vector at any time, with no limits on how the data is used or shared
- European publisher hosting on Google Cloud within the EU, with a detailed GDPR policy and retention periods stated purpose by purpose
Cons
- No pricing is published anywhere, and a pricing page that once existed has been removed: /pricing still redirects to a URL that now returns a 404
- The supply chain page claims that pricing is transparent and affordable, which contradicts the complete absence of any published figure
- No Data Processing Agreement is published or offered, and subprocessors appear only as categories with a single named provider
- No security certification is claimed anywhere on the site, neither SOC 2 nor ISO 27001
- No documented way for a customer to exclude their data from model training, and the vendor never states its position on the question
- Access is restricted to legal entities and professionals, so individuals and very small operators are excluded by the terms
- English only, with no mobile app and no comparison or alternatives page to help a buyer position the tool against competitors
Pricing & Plans
No pricing is published. Kanop sells by quotation: the terms and conditions state that the prices of the services subscribed to are set out in the Quotation, which is prepared from the projects the customer imports into the application and must be accepted in writing within thirty days. There is a free trial period, during which an account holder may import nature-based project data into the application at no charge, but no permanent free plan is described. Subscription arrangements are possible, with any period started due in full. Billing and payment terms are likewise specified in the quotation. Two figures in the terms should not be mistaken for prices: the fixed indemnity of 40 euros is a statutory late-payment collection charge, and the liability cap refers to sums received over the preceding twelve months.
Data, GDPR & hosting
A consolidated view of how Kanop handles your data.
GDPR overview
Implementation is concrete and specific. The privacy policy, in force since 24 October 2022, names RAIL SAS as controller, giving its registration with the Evry Trade and Companies Registry under number 899 308 878 and its head office at 3 rue Joliot-Curie, 91190 Gif-sur-Yvette. It cites Regulation (EU) 2016/679 explicitly, sets out purposes with their legal bases and a retention period for each, and lists ten data subject rights with their article numbers, including the right to complain to the CNIL under Article 77. Transfers outside the EU are covered by adequacy decisions, standard contractual clauses, binding corporate rules or certification. Data protection contact is privacy@kanop.io. No data protection officer is named, no Data Processing Agreement is published, and subprocessors appear only as categories. Being established in France, the publisher needs no Article 27 representative.
Who owns the data?
The terms draw a clear line. Kanop owns the application itself, along with the software, infrastructure, databases and content behind it; customers receive a non-exclusive, non-transferable SaaS licence that carries no transfer of ownership. The data a customer uploads and the reports generated from it remain under the customer's control: they can be exported at any time as CSV, raster or vector files, and Kanop states there are no restrictions on how that data is used or shared. One exception applies in the vendor's favour: Kanop data products may not be used to train artificial intelligence or machine learning models. Personal data is processed by RAIL SAS as controller, with staff, subcontractors and, where legally required, public authorities as recipients.
Reuse rights
Customers can reuse their own data freely without asking permission. Kanop states that uploaded data and generated reports can be exported at any time in the format of the customer's choice, and that there are no restrictions on how the data may be utilised or shared. The single stated limit is that Kanop data products may not be used to train artificial intelligence or machine learning models. Beyond that, the outputs are meant to travel: they are designed to feed directly into EUDR due diligence statements, GHG Protocol LSRS inventories, SBTi FLAG target tracking and TNFD or CSRD disclosures, and to be handed to auditors and third-party verifiers. On the vendor's side, the models are trained on third-party LiDAR reference data rather than customer data, although customer field measurements may optionally be used to recalibrate model outputs for that customer's own project.
Data retention & training
Hosting summary
Personal data is hosted for the duration of processing on Google Cloud Platform servers located in the European Union, as stated in the privacy policy. The terms repeat the commitment for the application and the data produced and entered on it, promising hosting on Kanop's own servers or through a professional hosting provider, on servers located in a territory of the European Union. No individual country and no cloud region is named, so the guarantee is regional rather than national. Transfers outside the EU may still occur through the third-party tools in use, such as the CRM, chatbot and mailing providers; these are framed by an adequacy decision under Article 45, by appropriate safeguards under Article 46 including standard contractual clauses, binding corporate rules or an approved certification mechanism, or by another safeguard under Chapter V of the GDPR. The publisher commits to protecting the infrastructure and detecting malicious acts as an obligation of means. No hosting certification, no data residency option and no security standard is claimed anywhere on the site.
Things to keep in mind
Risks and trade-offs to weigh before adopting Kanop.
- The client logos on the homepage belong to Kanop's customers and the 'They Support Us' band to standards bodies and memberships such as Verra, TNFD and the Biodiversity Credit Alliance. Neither is an investor list, a certification of the publisher, or an endorsement of any particular result.
- Verra, Gold Standard, Isometric, Equitable Earth, EUDR and CSRD are what the software helps you comply with, or credentials awarded to its data products. They are not security or quality certifications of the company itself, and the Equitable Earth benchmark PDF is a third party's document.
- The site claims transparent and affordable pricing while publishing no figures at all, and the former pricing page now returns a 404. Budget expectations cannot be formed before a sales conversation, which makes early comparison against alternatives difficult.
- Satellite estimates are statistical, not measurements of individual trees. Pixel-level error is roughly double site-level error, so reading a single 30 metre pixel as ground truth for a small parcel will mislead. Always carry the confidence interval into whatever decision follows.
- Audit-ready output is not the same as an audit. Kanop supplies the data layer; eligibility, validation and credit issuance remain with the standard bodies and verifiers, and a compliance claim still needs its own review.
- With no published DPA, no named subprocessor list and no security certification, a procurement or legal review will have to obtain those commitments contractually rather than from the website.
- The privacy policy dates from October 2022 and predates the AI agents and the MCP server. Routing project data through a third-party assistant such as Claude, ChatGPT or Le Chat introduces a processor that the published policy does not describe.
Setup & Integrations
Technical difficulty
Low to moderate, depending on the route. Sign-up is a form and the account opens automatically; you then import project boundaries as GeoJSON or shapefile, routine work for any GIS user. Nothing is installed, since the product is pure SaaS, and the AI assistant route needs no setup at all. The API is the most demanding: generate an application token in the web app, then integrate the REST endpoints, working from a public OpenAPI specification. The real prerequisite is domain knowledge, though the agents guide methodology selection and expert support is available.
Deployment
Integrations
Supported languages
Behind Kanop
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
How much does Kanop cost?
Is Kanop compatible with Verra, Gold Standard and other carbon standards?
Do I need field data to use it?
How accurate are the biomass estimates?
Is there an API, and is the documentation public?
Can I use Kanop from an AI assistant?
Who controls the data I upload?
Where is the data hosted?
Which reporting frameworks does it cover?
How quickly do results arrive?
Should you pick Kanop?
Kanop occupies a narrow, demanding niche and occupies it credibly. Its strongest argument is external validation rather than marketing: Verra vetting for VM0047 dynamic baselines, selection by Isometric across both partner categories, and a top ranking in Equitable Earth's independent biomass benchmark are judgements made by the standards bodies whose approval its customers actually need. The published accuracy figures reinforce that impression, since few vendors in this market disclose an RMSE, a validation protocol and a pixel-level error visibly worse than the site-level one.
The technical openness is unusual too. Eight OpenAPI specifications are readable without authentication, and the MCP server means an analyst can query forest carbon data from an assistant they already use, without opening a platform at all.
Against that, the commercial opacity is total. No price of any kind is published. A pricing page existed and has been removed, yet the supply chain page still claims that pricing is transparent and affordable. Every buyer goes through a demo and a quotation. Data governance is reasonable but incomplete: hosting is on Google Cloud in the European Union, retention periods are set out purpose by purpose, and customers can export everything at any time, but there is no published DPA, no named subprocessor list, no security certification and no documented way to opt out of model training. The privacy policy has not been revised since October 2022, before the AI agents and the MCP server existed.
Kanop suits organisations rather than individuals, since the terms restrict access to legal entities and professionals. For a carbon project developer, an investor screening a portfolio, or a corporate facing EUDR and LSRS deadlines, it is a serious candidate. For anyone who needs to compare prices before talking to sales, it is not.
- Choosing a selection results in a full page refresh.
- Opens in a new window.