Skip to main content

Η σελίδα υπάρχει και στα ελληνικά. Διαβάστε τη στα ελληνικά

Book an X-RayX-Ray Book a Workflow X-Ray: a fixed-price diagnosis

AI & Custom Solutions

Four ways to get hours back, a written standard for how we do it, and a diagnosis that tells you where to start.

What AI does inside a business: the overview

Services

Sectors

Not sure where to start?

Workflow X-Ray

In one to two weeks: a map of hours and euros, five opportunities with the payback worked out, and a demo running on your own data.

From €890 · credited 100% against the pilot

What we deliver, day by day

Strategy Lab · B2B manufacturing

Eight defects. Sixty seconds. Human vs Vision AI: a B2B marketing game for manufacturers

AI-generated image: an olive-oil bottling line with glass bottles on a stainless-steel conveyor, with a 60-second timer in the corner.This image was created with an AI tool (AI Act, Article 50).
60″ or untimed · mouse, touch or keyboard · no sign-up
Arrows: move · + and −: zoom · Enter: mark · Esc: finish

The AI-generated image shows an olive-oil bottling line. The eight defects are:

  1. Missing cap: An unsealed bottle: it leaks in the case and can't be sold.
  2. Upside-down label: The oil is fine; how it looks on the shelf isn't.
  3. Torn label: Information that must be legible is missing, and the shopper leaves it on the shelf.
  4. Label applied too low: A sign the labeller has drifted; left alone, it keeps happening.
  5. Underfilled bottle: Less product than the label promises: trouble with customers and with market inspectors.
  6. Foreign particle: The one you never want a consumer to find.
  7. Wrong cap: The wrong closure at the capper: it can mean a mixed-up batch or SKU.
  8. Tipped-over bottle: It can jam the line and knock over the ones behind it.

The machine: 8/8 · missed: 0 · false alarms: 3.

  • Label missing: The right-hand lane normally runs unlabelled. The model doesn't know the line's spec.
  • Label missing: The right-hand lane normally runs unlabelled. The model doesn't know the line's spec.
  • Dark spot on glass: Scuffing and reflection on the steel guide rail, not on the bottle.

An AI-generated image of an olive-oil bottling line hides eight deliberate defects. Find as many as you can. When the clock stops, you see what a vision model found in the same image.

It's a campaign and proof in one. A campaign, because no engineer likes losing to a colleague. Proof, because at the end they see with their own eyes what a vision model spots, and where it goes wrong.

Sector
B2B manufacturing
Mechanics
A few minutes of play that is the value proposition itself.You see the result on you or in your home before you buy.
Demo engine
Deep-Zoom
Complexity
1 = configuring existing tools · 5 = custom R&D
Production cost
Netex production only, excluding media and data licences. € under €2,000 · €€ €2,000–6,000 · €€€ €6,000–15,000 · €€€€ €15,000–40,000 · €€€€€ over €40,000. Netex estimate.
How this mechanism runs for a business in this sector, step by step.What a real model saw, mistakes included.

Decode

How it got you

Human vs Vision AIPlay → Decode → Deploy
Mechanism
Gm · Snackable game + Sv · Self-visualisation The engineer doesn't read about machine vision. They watch it work on a line like their own.
Psychology
Curiosity lights up the brain's reward circuit and sharpens memory, even for things seen while searching (Gruber, Gelman & Ranganath, 2014). First a test of skill, then a humbling moment: "I didn't know a machine could do that." That is our whole message, delivered as an experience. Gruber, Gelman & Ranganath (2014), Neuron (opens in a new tab)
Platform signal
LinkedIn: engineers send their score to colleagues, and the company board travels round the firm. The time is spent on your site, not in someone else's feed.
Honest range
2–5% of players from target accounts ask for an assessment · Sales cycle into an AI project: 3–6 months. Typical range for a pilot in a Greek SME · Netex estimate
I want one of these for my brand

Deploy

How we'd build it for you

Who it's forFor manufacturers who sell to manufacturers: packaging, machinery, materials, quality services. Your quality becomes something the buyer plays, not something they skim in a brochure.

  1. We photograph your line at very high resolution and stage real defects on it. Or we make the image with AI, with deliberate defects and the "AI-generated image" label, as in this demo.

  2. Engineers and plant managers get 60 seconds, with a company leaderboard on LinkedIn.

  3. At the end, the vision model shows what it found, with boxes and confidence scores, plus what it missed and what it got wrong.

  4. The button doesn't say "contact us". It asks for 20 defect photos from the player's own line, for a feasibility check.

Channels and tech

Channels
LinkedIn, organic and paid by job title and industry (3.1M members in Greece, DataReportal 2026) · email to industrial lists · a QR code on the trade-fair stand · a tool in the sales team's hands
Tech
Deep-zoom tiles (OpenSeadragon), a hotspot file, a timer with an untimed option, a leaderboard validated on the server, and the vision model's findings on the image, mistakes included.
Complexity
3/5
Production cost
€€–€€€ · Netex estimate

What to expect, honestly

We don't promise viral. We promise a mechanism you can measure.

  • Players from target accounts who ask for an assessment 2–5%
  • Sales cycle into an AI project 3–6 months

most likely on a first run · A first run usually lands in the lower half of the range.

Before launch we fix one primary KPI and a control: a holdout area, off-trigger days or the previous period.

Guardrails

  • Photos only with the plant's written permission, with no workers' faces and nothing commercially sensitive in frame.
  • The board shows real plays only. Never fake scores to "warm it up".
  • Lab accuracy isn't line accuracy. The game promises no percentages.
  • Sharing is optional, never a condition of entry.
  • The machine's boxes are real model output, mistakes included. Never a staged score, and never the finds without the false alarms beside them.
  • An AI-generated image always carries the "AI-generated image" label and is never presented as a photo of a real line.AI Act, Article 50

Who did it before us

Hidden-object games hold people for minutes, not seconds. They even work on serious B2B buyers, when the game is the value proposition itself.

Third-party results are mostly self-reported, from award entries and vendors. They show a mechanic can work, not what it will do for you.

Year Who What they did Result
2024 · B2B, international Frontify «Spot the Brand» 50 logos hidden in an illustrated city, five minutes on the clock. 40,000+ plays in the first month, +96% sales leads in two weeks self-reported source (opens in a new tab)
2021–23 · UK, IE, AU, NZ Cadbury «Worldwide Hide» Hide a virtual egg on Street View for someone you love. ≈5 minutes average time on site self-reported source (opens in a new tab)

From marketing to the line

What you just played is quality control.

The same technique, on cameras you may already have, looks at every part, not a sample. You find the defect, not your customer.

Machine vision is won or lost on lighting and training photos. That's why we start from your photos, not from someone else's percentages.

Send us 20 defect photos We'll tell you whether a camera can solve it before we talk about cost.

Or start with a Plant X-Ray, from €4,900.

  1. Camera on the line
  2. Model output with confidence
  3. Confidence threshold
  4. Human review queue for the grey zone
  5. Logging and retraining

Mechanism Sprint

Want this game built on your own line?

Mechanism Sprint: designed, produced and measured in 10 working days. From €1,500 plus production.

I want one of these for my brand The whole Strategy Lab
We reply within one working day.

The scenario you just played goes straight into the form.