Spheer App
Spheer App is a browser-based geospatial monitoring platform. Users draw or import a handful of labelled observations, and a proprietary Sentinel-2 foundation model turns them into prediction maps covering entire regions within minutes, exportable to standard GIS tools.
What is Spheer App?
Spheer App is a cloud platform that turns satellite imagery into custom monitoring maps, built by Spheer in Groningen, the Netherlands. Until early 2026 the product was called Carto; it was renamed Spheer to align the app with the company name.
The working principle is deliberately narrow. Instead of asking users to build a model, the app asks them for examples. You draw polygons over your area of interest, or import them, and label what they show. Spheer combines those few observations with its own geospatial foundation model and returns a prediction map covering the whole area. Because the underlying model already understands the earth's surface, the vendor claims roughly a hundred times fewer examples are needed than a conventional approach would require, and a usable map in about ten minutes rather than weeks.
The data behind it is Sentinel-2, the European satellite constellation that images the globe every five days at resolutions down to ten metres, with usable history back to 2017. Rather than reading a single image, Spheer analyses a full year of imagery so that seasonal variation can be told apart from structural change. The foundation model itself, released as Albatross-NL and Albatross-EU in June 2026, is a proprietary multilayer Transformer trained by self-supervision on roughly 250 GB of unlabelled imagery, capable of semantic segmentation, object detection and temporal change recognition.
Six application areas are advertised: nature, water, agriculture, spatial planning, climate adaptation and ESG. Work is organised into projects, areas of interest in which one or more indicators are followed from year to year, and several organisations can collaborate on a shared picture of the same territory. Everything runs in the browser; nothing is installed or administered by the customer. Output leaves in common formats for QGIS, ArcGIS or other spatial workflows, and an API is offered for deeper integration.
Spheer reports more than 150 active users across 20 organisations, 2.58 million hectares under monitoring and over 400 completed cases. Named customers include the provinces of Groningen and Noord-Brabant, the water company PWN, and BIJ12, which completed an independent validation of the tool.
What it does
- Turn a handful of drawn observations into prediction maps covering an entire region
- Monitor vegetation, biodiversity, water quality, crops and land use over time
- Compare the current situation with historical change going back to 2017
- Detect change across a landscape without supplying examples
- Measure model quality with a built-in confusion matrix
- Export maps and analyses to QGIS, ArcGIS and other GIS workflows
- Share projects, models and insights with colleagues and partner organisations
When to use Spheer App / When not to
A quick filter to help you decide if Spheer App is the right fit.
When to use Spheer App
- Provincial and municipal authorities that must survey large territories repeatedly and report on them
- Nature and water managers tracking vegetation, biodiversity or water quality year after year
- Teams with strong field expertise but no data scientist or in-house AI capability
- GIS departments already working in QGIS or ArcGIS that want to feed existing spatial workflows
- Organisations checking agricultural measures or land-use compliance across thousands of hectares
When not to use Spheer App
- Life-critical uses the vendor explicitly excludes: military, autonomous navigation, disaster response without human validation
- Anyone needing real-time operation or automated decision-making without human oversight
- Individuals: accounts are issued through client organisations, with no self-service sign-up
- Projects mapping features smaller than roughly ten metres across, below Sentinel-2 resolution
- Buyers who need published contractual terms, a DPA or a documented hosting jurisdiction before signing
How to use Spheer App
A typical end-to-end flow, from setup to results.
- Obtain an account through your organisation; Spheer issues a login link by email
- Set a password on first sign-in, then work from app.spheer.ai in any browser
- Create a project and define the area of interest you want to follow
- Draw observations directly on the map, or import existing polygons
- Label what each observation shows, keeping examples precise rather than numerous
- Include an 'other' class and vary your examples so the model can discriminate
- Draw observations across at least two different years to capture growth patterns
- Train the model and wait a few minutes for predictions across the whole area
- Review the maps, add observations where predictions are weak, and retrain
- Export the finished maps to QGIS, ArcGIS or another spatial workflow
Pros & Cons
Pros
- Usable maps in about ten minutes where conventional mapping takes weeks
- Very few labelled examples needed, thanks to the pre-trained foundation model
- No AI or remote-sensing skills required; domain expertise is enough
- Scales from a single nature reserve to an entire province and beyond
- Fits existing GIS workflows through standard export formats rather than replacing them
- Unusually candid AI Act transparency: published model cards, stated risk class, documented limitations
- Verifiable public references, including an independent validation completed by BIJ12
Cons
- No public pricing whatsoever: no plans, no ranges, everything goes through sales
- No self-service access and no announced free plan or trial; accounts come via a client organisation
- Very thin legal documentation: a one-sentence privacy policy, no public terms, no DPA, no subprocessor list
- The hosting jurisdiction is never stated and no security certification is claimed
- An API is sold and cited in the model cards but has no public documentation
- A single data source at ten-metre resolution: fine features are out of reach and other sensors are untested
- Accuracy is demonstrated only for the Netherlands and parts of Europe, and degrades under heavy cloud cover
Pricing & Plans
Spheer does not publish any pricing for the Spheer App. There is no pricing page, and a review of the complete sitemap confirms that none exists. No permanent free plan and no free trial are announced, and access requires a user account issued through a client organisation. Prospective customers must request a demonstration or a quotation by contacting the vendor directly. The only monetary figure on the site, 15,000 EUR, appears in a customer testimonial describing what conventional mapping normally costs that organisation; it is not a Spheer price.
Data, GDPR & hosting
A consolidated view of how Spheer App handles your data.
GDPR overview
Spheer is established in the Netherlands, so no Article 27 representative applies. A help-centre entry asks whether the app complies with the AVG, the Dutch implementation of the GDPR, and answers that personal data in the app is limited to login names and email addresses, invisible outside it; that user activity is logged; and that this information can be handed over or deleted on request. That is the whole of it. The privacy policy is one sentence about confidentiality. There is no stated lawful basis, no retention period, no list of data-subject rights, no DPO, no public terms of use, no data-processing agreement, no subprocessor list and no security certification. AI Act compliance, by contrast, is documented in detail through published model cards.
Who owns the data?
The published privacy statement is a single sentence: everything a user supplies to Spheer is treated as confidential and is not disclosed to third parties unless the law or the judicial authorities require it. The product page adds that customers keep control of their own datasets and models inside the platform. The intellectual-property clause in the disclaimer covers the website itself, its texts, images, logos and software, and not customer content. Beyond that, ownership is undefined: no public terms of use, no licence clause on uploaded observations and no data-processing agreement are published, so contractual ownership has to be settled directly with the vendor.
Reuse rights
No published terms grant or restrict reuse, because Spheer publishes no terms of use at all, although they do exist since the model cards quote them verbatim. In practice the platform is built around reuse by the customer: maps and analyses export in common formats and can be taken into QGIS, ArcGIS or other spatial workflows, models and insights can be shared with colleagues and partner organisations, and an API is offered for deeper coupling. What Spheer itself may do with customer observations is not addressed anywhere. Anyone needing a firm answer on redistribution rights must request the terms of use from the vendor.
Data retention & training
Hosting summary
Spheer publishes nothing about where customer data is hosted. A targeted search across every page and document collected found no mention of a hosting country, region, cloud provider or data centre, and there is no trust page, no subprocessor list and no data-processing agreement to consult. The app is described only as a fully cloud-based service that the customer does not have to administer. Two facts are known but must not be mistaken for a hosting commitment. The vendor is established in Groningen, the Netherlands, and is therefore subject to EU law. Separately, the public website resolves to an IP address located in Amsterdam and operated by DigitalOcean, which describes where the marketing site is served, not where user observations, trained models or account data are stored. For any buyer with data-residency obligations, the hosting jurisdiction is an open question that has to be put to the vendor directly.
Things to keep in mind
Risks and trade-offs to weigh before adopting Spheer App.
- Maps look authoritative, yet the vendor insists results must never be read without ground truth and context
- Automated decision-making is explicitly excluded, but convincing outputs invite exactly that shortcut
- Deskilling risk: teams that stop sending surveyors into the field lose the ability to judge when the model is wrong
- Contractual terms are not public, so data ownership and reuse rights stay unresolved until you obtain them
- No stated position on whether your observations are used to train the vendor's own models
- The hosting jurisdiction is undocumented, which is awkward for public-sector procurement
- An API is sold without documentation; do not architect around it before confirming its scope
Setup & Integrations
Technical difficulty
Setup is essentially frictionless: there is nothing to install, a browser is enough, and the first sign-in consists of following an emailed link and choosing a password. The real effort is methodological rather than technical. Model quality depends on how observations are drawn, and Spheer publishes guidance on it: favour accuracy over volume, split ambiguous areas, include an 'other' class, spread examples across at least two years, and keep each observation between 10 m and 1 km across. Expect an iterative loop of training, reviewing and adding examples. A user manual and an interpretation guide are provided.
Deployment
Integrations
Supported languages
Behind Spheer App
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does the Spheer App actually do?
Do I need technical skills to use it?
Which satellite data does it use, and how far back?
How large an area can be monitored?
How much does it cost?
How do I get access to the app?
Can I connect it to my existing GIS?
Is it compliant with the EU AI Act?
What limitations does the vendor acknowledge?
How is my personal data handled?
Should you pick Spheer App?
Spheer App is a focused, technically credible product. Few companies of this size train their own geospatial foundation model, and fewer still publish model cards stating the training set, the compute budget, the risk class under the EU AI Act and the conditions in which accuracy degrades. The public references are checkable, including Dutch provinces, the water company PWN and an independent validation completed by BIJ12, and the core promise is specific enough to test: a usable map in minutes from a handful of examples, at a scale conventional mapping cannot match economically.
The reservations are almost entirely commercial and contractual rather than technical. Nothing about price is public, there is no self-service route in, and the legal documentation falls well short of what a public buyer normally requires: a privacy policy of one sentence, no published terms of use, no data-processing agreement, no subprocessor list, no stated hosting jurisdiction and no security certification. Terms of use do exist, since the model cards quote them, but you have to ask for them. The vendor also takes no public position on whether customer observations feed its own models, which is worth settling before signing.
Scope matters too. The models are trained on the Netherlands and parts of Europe, the imagery is Sentinel-2 at ten metres, and Spheer is explicit that accuracy may fall outside those conditions and that human oversight is required throughout. Signs of a young company remain visible: the rebrand from Carto is incomplete, and one page still carries placeholder text.
For a Dutch or European public body already doing repetitive landscape monitoring, this is a serious candidate that deserves a demonstration. Treat the contractual and data-governance questions as the real due diligence, because the website will not answer them.
- Choosing a selection results in a full page refresh.
- Opens in a new window.