Scortex: our review
Scortex makes Spark, an AI vision unit installed at the end of a production line to detect surface defects on parts as they pass by, without sending the inspected images outside the factory. The vendor, based in Paris, has belonged to the Trigo group since 2022.
What we take away from it
What counts in its favor
- Inspection runs on a computing unit installed on the line: production images don't travel to a remote server for the accept-or-reject decision on a part.
- The model learns from conforming parts rather than from fixed, hand-written rules, which makes it possible to catch micro-defects that are hard to describe in a traditional spec sheet.
- 12 software updates shipped in 2025 across more than 25 million parts inspected, a sign of a product that keeps evolving rather than a stalled project.
What counts against it
- No public pricing page: the price is negotiated via a quote, and the vendor publishes no ballpark figure beyond a generic ROI calculator.
- No information security certification (ISO 27001, SOC 2) is mentioned anywhere on the site, even though the product inspects parts that may fall under the client's industrial secrets.
- The published privacy policy covers only the personal data of site visitors; it says nothing about the processing of the production images themselves, which is left entirely to the commercial contract.
A sound choice for an industrial site that wants to keep the inspection decision on site, provided the contract spells out, in black and white, what the public documentation doesn't say about production images.
The facts
- Data hosting
- Hosting not verified
- Categories
- Manufacturing
- Data verified on
- 30 July 2026
What happens to your data
- Is your data used to train their models?
- Subprocessors
- Not published. The privacy policy mentions a "data host" without naming it, and does not distinguish personal site data from industrial inspection data.
- Retention
- For the site's personal data: three years without interaction. Nothing is published on the retention period for the production images captured by Spark.
Scortex in detail
Scortex sells Spark, a visual inspection system that replaces or complements an operator's eye at the end of a production line. A camera, lighting and a compute unit watch each part as it passes and decide whether it's scrapped or shipped. The publisher is based in Paris, founded in 2016, and has belonged since 2022 to the Trigo group, a specialist in industrial quality control.
What it actually does
A computer vision model first learns what a compliant part looks like, from photos of parts that are. Once trained, it detects deviations: scratches, stains, a missing part, a mirage defect on a polished surface. The kit sold combines cameras, lighting, a touchscreen and a compute unit physically installed on the line — not simply access to a remote API. A dashboard, the Quality Center, then aggregates the results and remains accessible remotely for production monitoring. As of 2025, the publisher claims more than 25 million parts inspected across four continents, in cosmetics, automotive and electronics.
Why it interests us
This is the clear-cut example of the question that matters on a factory floor: where the decision gets made. At Scortex, it's made on the unit installed on the line, not in a remote data centre — a production line can't stop because an internet link drops. This is a genuine use of AI, not a marketing dress-up of classic control software: the model learns a profile of "normality" from the client's own compliant parts, which a fixed-threshold system doesn't do. See also our page on AI for industry.
Where it falls short
Three reservations before signing. First, price: no pricing page, only a quote after a sales contact — expected in industrial B2B sales, but it prevents any quick comparison between providers. Second, security: neither ISO 27001, nor SOC 2, nor any information security certification appears on the site, even though the inspected images can reveal manufacturing details the client considers confidential. Finally, the published privacy policy only covers website visitors' personal data: it says nothing about what Scortex does with images of industrial parts, nor about who can access them remotely for maintenance. This point gets negotiated in the contract, not on the website.
Our recommendation
Worth considering for a site that wants to keep the inspection decision local and doesn't need a public price list to start a discussion. Before signing, get in writing what the privacy page doesn't say: how long production images are retained, remote access by technical support, and any security certification the publisher can claim beyond its public website.
Who it's for, and who it isn't
Suited for
- A series-production site (cosmetics, automotive, electronics, packaging) that wants to detect surface defects without depending on a stable internet connection for every decision
- A manufacturer that refuses, as a matter of principle, to let photos of its parts leave its network to be judged elsewhere
Not a fit
- Small-batch or prototyping work, where volume doesn't justify training a model on conforming parts
- A buyer who needs a listed price to compare quotes before even a first sales contact
Frequently asked questions
- Do photos of my parts leave my factory?
- For the accept-or-reject decision, no: that decision is made by the unit installed on the line. A remotely accessible dashboard then aggregates the results, and the vendor can connect remotely for maintenance — exactly what that access covers isn't detailed publicly, so have it spelled out in the contract.
- How much does Spark cost?
- No price is published. Scortex works by quote, offers an online ROI calculator, and directs you to a sales contact for any actual figure.
- Does inspection require a permanent internet connection to work?
- Not for the inspection itself, which runs locally on the unit. A connection is still useful for the monitoring dashboard and updates, but the vendor doesn't specify what happens during an extended outage.
- Is Scortex still an active product, or an abandoned project?
- Active: the Trigo group, which acquired it in 2022, was still presenting a new product (Spark Multi View) and a trade show planned for 2026 at the time this listing was checked.
- What's the difference with traditional quality-control software?
- Traditional quality control compares the part to fixed rules (dimensions, thresholds). Scortex builds its model from conforming parts and then spots deviations, including defects that are hard to turn into rules — at the cost of a learning phase on real parts before going live.
Our sources
- Spark product pageaccessed on 30 July 2026
- Privacy policyaccessed on 30 July 2026
- Legal noticeaccessed on 30 July 2026
- About pageaccessed on 30 July 2026
- ROI calculatoraccessed on 30 July 2026
Prices and hosting terms change often. We recheck them periodically, but the publisher's page remains the only up-to-date reference.