Edge Scan
Edge Scan is an industrial computer vision system from the Polish company Edge AI. Installed on a concrete paving stone line, it inspects 100% of production in real time, measuring height to ±0.5 mm and flagging holes and inclusions.
What is Edge Scan?
Edge Scan is an industrial computer vision system built by Edge AI, a Polish technology startup based in Katowice. It sits on a concrete paving stone line, on the wet side just after the vibropress, and inspects every piece that passes underneath in real time. The pitch Edge AI makes is blunt: a worker checks around 3% of production, Edge Scan inspects 100%.
The system targets three problems the company sets out. First, the height standard: because you cannot measure every stone by hand, plants keep a safety buffer and pay for excess material on every single piece. Second, quality control: defect counts were usually checked about once an hour, with little real effect on quality problems. Third, reaction time: a fault can run for hours before anyone catches it by hand.
It works on three measurements. For height, several thousand measurement points are recorded on every stone and a median derived from them, stated as accurate to 0.5 mm; each product can carry its own height standard and allowed deviations. Measuring across many points also yields the overall curvature of the product and of the production boards, surfaced on a heatmap before deviations become a problem. For defects, an AI model trained on hundreds of thousands of production board scans identifies holes and inclusions, including ones sharing the colour and texture of the base material, and marks each one with a confidence score.
Results feed the Edge Scan web app, which offers dashboards, batches, analytics, products, operators and shifts, together with per-shift reports, in place of spreadsheets and manual reporting. Output also goes to plant systems over Modbus TCP or Profinet.
The published specification is unusually complete: 1400 mm working width, 10 to 600 mm inspected product height, a 4508 × ~1000 px height map per board, ±0.5 mm height accuracy, a 0.2 mm smallest detectable defect, three seconds to analyse a production board, and a six to eight week lead time. Installation and calibration are announced at two to four hours, with no change required to existing automation. The gains claimed, 97% fewer complaints and 3-5% material savings, are the vendor's own figures, with no Edge Scan customer named in support.
What it does
- Inspect 100% of production in real time instead of a manual sample
- Measure the height of every piece to ±0.5 mm from several thousand points
- Detect holes and inclusions with an AI model, scoring the confidence of each find
- Reveal product and production board curvature on a heatmap
- Push results to the plant PLC or MES over Modbus TCP or Profinet
- Track quality by batch, product, operator and shift in a web app
- Set a height standard and tolerated deviations for each product
When to use Edge Scan / When not to
A quick filter to help you decide if Edge Scan is the right fit.
When to use Edge Scan
- Paving stone and concrete slab manufacturers running vibropress lines who currently inspect only a small sample of their output
- Quality managers and inspectors who want every piece measured rather than an hourly manual spot check
- Production and process engineers chasing the material buffer kept on every stone to stay safely within height spec
- Plant automation teams whose PLC or MES already speaks Modbus TCP or Profinet
- Operations managers who want quality reporting by batch, operator and shift in a web app instead of spreadsheets
When not to use Edge Scan
- Buyers looking for a pure software product: this is a machine installed on a line, with a stated six to eight week lead time
- Manufacturers of anything other than concrete paving products, since the announced scope is paving stones and slabs
- Lines outside the published physical envelope of 1400 mm working width and 10 to 600 mm product height
- Teams that need published pricing, self-service sign-up or an online trial before speaking to a salesperson
- Anyone who needs a mobile app or a documented product API: neither exists, and integration runs over industrial protocols
How to use Edge Scan
A typical end-to-end flow, from setup to results.
- Get in touch through the form on the Edge Scan page, or by email or telephone
- Discuss your line, your product range and the quality problems you want solved
- Have Edge AI analyse your production line and confirm it fits the working envelope
- Agree the order, allowing for the stated six to eight week lead time
- Have the unit mounted on the wet side of the vibropress line, with installation and calibration announced at two to four hours
- Integrate the output with your automation and systems over Modbus TCP or Profinet, with no change to existing automation
- Set the height standard and the tolerated deviations for each product in your catalogue
- Run the production tests and have your team trained by Edge AI
- Work alongside Edge AI for the first few weeks while the system is calibrated to your product range and your line
- Use the web app day to day for dashboards, batches, analytics, products, operators and per-shift reports
Pros & Cons
Pros
- A full technical specification is published in figures, so a plant can judge the fit before contacting anyone
- Every piece is inspected, rather than the small sample a manual check allows
- Announced as working with a line from any manufacturer, with no change to existing automation
- Installation and calibration announced at two to four hours
- Standard industrial protocols, Modbus TCP and Profinet, rather than a proprietary link
- A quality analytics web app is part of the product, with reporting by batch, operator and shift
- Unusually transparent publisher: postal address, VAT number, KRS number and named founders with their backgrounds
Cons
- No pricing whatsoever is published: no figure, no range, no order of magnitude
- No terms and conditions and no contract template appear anywhere on the site
- The headline gains, 97% fewer complaints and 3-5% material savings, are vendor claims with no named Edge Scan customer or study behind them
- The privacy policy covers only the website; nothing is published on the ownership, retention or model training of the production data the system captures
- No DPA is published or offered, and no opt-out from training is documented
- Narrow scope and a hard physical envelope: concrete paving products, 1400 mm working width, 10 to 600 mm height
- A single personal email address serves the whole site, with no dedicated support or legal mailbox, from a team of around ten people
Pricing & Plans
No price is published. Edge AI operates no public price list, and none of the site's pages carries a figure, a range or a tier for Edge Scan; the system is quoted on request after contact with the company. There is no free plan and no free trial announced as such. Prospective buyers should treat this as industrial capital equipment rather than a subscription, with a stated lead time of six to eight weeks, detailed technical documentation and remote support being presented as shipping options. One caution: the only monetary amounts appearing on the site, PLN 16.7 million and PLN 16.1 million, are the budget of the IDEA Global acceleration programme and bear no relation to the price of Edge Scan.
Data, GDPR & hosting
A consolidated view of how Edge Scan handles your data.
GDPR overview
Implementation is concrete for the website and silent for the product. The privacy policy, last updated 29 August 2026, names Edge AI as controller, cites explicit legal bases, Art. 6(1)(f) for handling correspondence and Art. 6(1)(b) for pre-contractual steps, and lists data subject rights: access, rectification, erasure, restriction, portability and objection. It names the supervisory authority, the Polish Prezes Urzędu Ochrony Danych Osobowych in Warsaw, and states that some providers process data outside the European Economic Area under standard contractual clauses approved by the European Commission. Subprocessors are named. No DPO is appointed, and no Article 27 representative is required since the company is established in Poland. No DPA is published or offered, and the policy says nothing about the production data captured on customer lines.
Who owns the data?
Edge AI, at ul. Porcelanowa 25 in Katowice, names itself the controller of the personal data collected through edgeai.pl. That published scope is narrow: it covers the contact form, visit statistics and hosting logs, and nothing more. The company publishes no terms and conditions and no contract template, so nothing on the site states who owns the images, height maps and defect records that Edge Scan captures on a customer's production line, nor what Edge AI may do with them. Ownership of production data is therefore something to settle contractually before deployment, not something the public documentation answers.
Reuse rights
For website data the stated purposes are narrow and legally grounded. Form submissions, meaning an email address and the message body, are used to answer the enquiry and, where the conversation goes that way, to agree the terms of working together. Edge AI cites legitimate interest in handling correspondence under Art. 6(1)(f) GDPR and pre-contractual steps under Art. 6(1)(b) GDPR. The form itself is handled by Web3Forms, which forwards messages to the company mailbox. Visit statistics come from Cloudflare Web Analytics, which the policy says stores no cookies, builds no profiles and does not track people across sites, collecting only aggregates such as visit counts, page paths, country and referrer. Edge AI states it uses neither Google Analytics nor any advertising tool, and keeps only the chosen language version in browser local storage. Nothing is published about the reuse of production data captured by Edge Scan, and no opt-out from model training is documented anywhere on the site.
Data retention & training
Hosting summary
The published hosting information concerns the website only. Edge AI states that edgeai.pl runs on Cloudflare Pages and that the host processes standard logs, including IP addresses, in order to serve the site and keep it secure. Visit statistics come from Cloudflare Web Analytics, and the contact form is handled by Web3Forms. The policy names its subprocessors, Cloudflare for hosting and statistics, Web3Forms for the form and the company's email provider, and states that some of them process data outside the European Economic Area under standard contractual clauses approved by the European Commission. No hosting country or region is ever named, which is why those fields are left empty here. The IP address the domain resolves to belongs to a Cloudflare anycast node and says nothing about where data actually sits. Above all, nothing is published about where the images, height maps and defect records captured by Edge Scan on a customer's production line are stored or processed.
Things to keep in mind
Risks and trade-offs to weigh before adopting Edge Scan.
- The performance figures, 97% fewer complaints and 3-5% material savings, come from Edge AI alone: no Edge Scan customer is named and no study is cited
- The five case studies on the site, including STS S.A., CAT LC Polska and GardianAI, are other Edge AI projects and not Edge Scan deployments; the REST API mentioned in one of them belongs to a client's system, not to Edge Scan
- Nothing published governs the images and measurements captured on your line, so ownership, retention and any use for model training must be pinned down in the contract
- With no public pricing at all, budgeting is impossible before a commercial conversation and comparison with alternatives is difficult
- Automating an inspection people used to perform can erode a team's hands-on judgement; a confidence score is a probability, not a verdict, and thresholds deserve periodic human review
- The publisher is a company of around ten people, registered very recently; for equipment that sits in a production line for years, support capacity and continuity are worth testing
- The trial slot announced on 27 August 2026 covers a single line and its financial terms are not stated: check whether it is still open and what it actually involves
Setup & Integrations
Technical difficulty
This is a physical deployment on a production line, not a software installation, but Edge AI carries most of the load: line analysis, mounting, integration with automation and systems, production testing and team training. Installation and calibration are announced at two to four hours, with no change required to existing automation and compatibility claimed with any manufacturer's line. The customer must supply a contact during commissioning, typically the process engineer or production manager, and should expect a few weeks of joint work while the system is calibrated to its product range. Lead time is six to eight weeks.
Deployment
Integrations
Supported languages
Behind Edge Scan
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does Edge Scan inspect?
How accurate is it?
What kind of defects does it find?
Does our production line have to be modified?
How does it connect to our PLC or MES?
What are the physical limits?
Is there software included?
What does it cost?
Who is behind Edge Scan?
What happens to the production data the system captures?
Should you pick Edge Scan?
Edge Scan is a narrow, serious product for a narrow, well-identified segment. Edge AI publishes what most industrial vendors keep back: a complete technical specification in figures, so a paving stone manufacturer can work out whether the system fits its line before speaking to anyone. The engineering story is coherent, from the height median over several thousand points to the defect model and the heatmaps, and the integration path runs over ordinary industrial protocols rather than a proprietary link.
The reservations are about documentation rather than technology. Nothing is priced, so a plant cannot form a budget or compare alternatives without a commercial conversation. The gains that headline the product page are the vendor's own figures, and the five case studies on the site belong to other Edge AI projects, not to Edge Scan deployments. Most importantly for a system that will photograph a customer's output continuously, no published document says who owns those images and measurements, how long they are kept, or whether they train anyone's models: the privacy policy stops at the website, and there are no terms and conditions at all.
The publisher is young, small and recently incorporated, but genuinely transparent about its identity, its founders and their credentials, and it has just secured public grant funding to install and test the system on further lines. For a manufacturer with the right press and the right products, Edge Scan is worth a conversation, provided that conversation covers the commercial terms and the fate of production data, since the website answers neither.
- Choosing a selection results in a full page refresh.
- Opens in a new window.