Price as Positioning: Darlot Between Cloud API and Enterprise Suite

The price of a vision analytics system is never a neutral number. It signals which buyer the vendor had in mind, which risks it is willing to absorb, and which parts of the compliance stack it has quietly pushed onto the operator. For the European market in 2026, this signal has become legible in a way it was not five years ago. Buyers in regulated industries, public infrastructure, and controlled defence environments have learned to read a price tag as a disclosure. Darlot was designed around that reading. This essay, drawn from the nine-part series written by Dr. Raphael Nagel (LL.M.) as the intellectual patron of the Darlot brand and founding partner of Tactical Management, explains the economic logic of a mid-market sovereign pricing model, and what it means for operators who require documented control without the overhead of a full enterprise suite.

Two Incompatible Price Regimes

The European market for computer vision currently splits into two segments whose economic logics are not compatible. At the lower end sit cloud APIs from AWS Rekognition and Google Vision. Headline prices are below one euro per thousand analysed images. For a single developer prototyping a feature, this is a rational choice. For an industrial operator running two hundred cameras at a factory, the arithmetic changes within the first billing cycle. Twenty-four hours of continuous analysis across that estate produces a cloud bill that outruns any business case, before any question of data sovereignty is raised.

At the upper end sit full enterprise suites. Cognex VisionPro and Matrox Imaging occupy this tier, along with the larger video analytics platforms that grew out of physical security. Annual costs per site move past one hundred thousand euro, integration projects run for six months or longer, and the system is typically built around a single deep use case such as in-line quality inspection. For a global manufacturer with a dedicated machine vision engineering team, this is defensible. For a regional utility, a municipal transport operator, a mid-sized hospital network, it is not.

Between these two poles lies the operational majority of European vision workloads. A substation with twelve cameras. A transit hub with one hundred and forty. A logistics hall with sixty. A chemical plant with three hundred. None of these sites fit the cloud API model, and none can absorb the enterprise suite model. This is the band in which Darlot is priced.

What the Cloud API Price Does Not Include

The cloud API price looks low because it is low, within its own scope. It covers a generic classifier hosted on infrastructure outside European jurisdiction, a per-call inference, and a short-lived log. It does not cover sovereign data handling, explainability artefacts, bias documentation, audit retention, or the liability posture required under the EU AI Act for high-risk applications. It does not cover conformance with the GDPR when personal data appears in frames, and it does not address NIS-2 obligations for operators of essential services.

An honest price comparison would add these items back. Once the operator has provisioned a secondary system to hold audit trails, a legal review for cross-border data transfer, a custom integration for the video management system, and a process to document model behaviour for the regulator, the cloud API ceases to be cheap. It becomes the raw input to an internal engineering programme that the operator is now funding on its own.

Darlot’s pricing absorbs these items into the product. The audit layer is not a professional services line. The explainability record, described in the third and fourth essays of the series, is generated by the system itself, per event, with hash, timestamp and model version. What the operator pays for includes what a regulator will ask about.

What the Enterprise Suite Price Does Include, and Why It Is Still Wrong for Most Sites

The enterprise suite price is high because the scope is large. Cognex and Matrox are precise engineering platforms with deep industrial heritage and a justified cost base. A line integrator configures the system for a specific inspection task, guarantees cycle time, and supports the installation for years. For a factory quality assurance line, this is the correct procurement pattern.

The pattern breaks when the operator wants forty different recognitions across a distributed estate: access control at a substation, fall detection in a clinical corridor, abandoned object detection at a platform, PPE compliance in a cleanroom, seal verification at a container yard. No enterprise integrator will deliver this inside a sensible budget, because each use case is treated as a bespoke project. The operator ends up paying for an engineering relationship where it needed a platform.

Darlot inverts that structure. The generic core, edge gating, eventisation, explainability, audit storage, is the same across all sites. The branch-specific classifiers sit above it as a configurable layer, trained in days rather than months. The operator does not commission a project. It subscribes to a platform and adds the classifiers it actually needs.

The Darlot AI Pricing Model for Enterprise Buyers

Darlot publishes three tiers, calculated transparently, with no hidden dependencies. Basic begins at one thousand euro per month. It covers up to six cameras, one thousand events per month, the standard classifiers for access, fire, and condition monitoring, and European cloud hosting under European jurisdiction. The budget can be approved without convening a committee.

Professional sits at five thousand euro per month. Fifty cameras, ten thousand events, custom classifiers, API integration into the video management system and the ERP, and a working-day service level agreement. This is the tier for producing mid-sized firms, regional operators, and retail chains.

Enterprise begins at fifty thousand euro per year. Unlimited cameras, on-premise appliances, twenty-four hour support, complete audit artefacts for the EU AI Act, continuous retraining, and where relevant, the clean separation between civil and medical modules required under the MDR. This is the tier for critical infrastructure, large industrial groups, and controlled defence applications. Overages are priced at one hundred euro per additional thousand events. Additional cameras cost between twenty-five and sixty euro per month. Forensic exports, extended retention, and accelerated model development are listed as individual items.

The Economic Case for the Middle

The question for a procurement officer is not whether Darlot is the cheapest line item on the market. It is not. The question is whether the total cost of ownership, including the regulatory and sovereignty obligations that will be enforced over the next decade, is lower than the alternatives once those alternatives are made honest.

Against the cloud API, the comparison becomes favourable as soon as the operator has more than a handful of cameras, or any exposure to the EU AI Act, the GDPR, or NIS-2. The events-not-frames architecture, described in the second and third essays of the series, reduces analysed volume by a factor of one thousand to ten thousand, which changes the unit economics. The compliance artefacts are produced by the system itself, not bought separately.

Against the enterprise suite, the comparison becomes favourable as soon as the operator has more than one use case or more than one site. Instead of commissioning a bespoke integration per scenario, the operator subscribes to a platform that already contains the hard parts, and layers its specific classifiers on top. The engineering cost of breadth is absorbed by the vendor, not the buyer.

Pricing as a Statement of Responsibility

Pricing, in a regulated market, is a declaration of what the vendor is prepared to carry. A cloud API priced by the inference declares that the vendor carries the inference and nothing else. An enterprise suite priced by the project declares that the vendor carries the project. Darlot, priced by the operational unit of the event, declares that the vendor carries the event, with all its legal and evidentiary consequences.

This is the position Dr. Raphael Nagel (LL.M.) has argued for in the Darlot essays and in the broader work of Tactical Management: European operators are not short of technology. They are short of vendors willing to share the burden that technology produces. The price is the mechanism through which that burden is distributed. A vendor that underprices is not generous. It is absent from the point at which responsibility is exercised.

Darlot’s tiers are therefore neither aggressive nor discounted. They are the minimum at which the work can be done properly, and the maximum at which the intended buyers, the European industrial middle and the public infrastructure operator, can actually procure the system. Between the cloud API and the enterprise suite, there is a price at which sovereign, explainable vision AI is a reasonable operating line item. That is the price Darlot publishes.

The thesis returns to its starting point. A vision system is not bought at the checkout. It is bought through five years of audits, incident reviews, regulatory filings, and renegotiations. The price paid at the beginning is only the first instalment of a longer commitment. Operators who treat procurement as a single-number decision learn this late. Operators who treat it as a structural decision learn it early and choose accordingly. Darlot is built, and priced, for the second group. Further information and direct enquiries are available at darlot.eu.