# AicobotiX — AI visual quality inspection (QiCHECK / QiPACK / QiSHELL)

> Set up by your own operators. Deployed without touching your line. Running in production today on furniture-component, power-electronics, enamelware and window-manufacturing lines across Central Europe. 98.9% live pass rate in production · ~120 ms per inspection, end to end · Operators train it — no vision engineers · Weeks to deploy, not quarters

AicobotiX s.r.o., Priemyselná 5/C, 917 01 Trnava, Slovakia. info@aicobotix.com · IČO 52271935 · DIČ 2120958884 · IČ DPH SK2120958884.
Site available in English and Slovak.

## Products
- [QiCHECK](https://www.aicobotix.com/products#qicheck): The flagship platform for assembly and PCB inspection: presence, completeness, correctness and cosmetic defects — on manual, semi-automated or robotic stations.
- [QiPACK](https://www.aicobotix.com/products#qipack): Watches every package on the conveyor. Your QC team defines what a good pack looks like; QiPACK checks seals, counts and labels on every unit and gives operators instant feedback.
- [QiSHELL](https://www.aicobotix.com/products#qishell): The box is sealed — is the right thing inside, and is the package itself flawless? QiSHELL verifies labels, codes and print quality, and inspects the closed package: correctly glued and sealed, no creases, no tears.
- [QiPICK](https://www.aicobotix.com/products#qipick): Vision-guided robot picking without a 3D scanner: neural networks trained per part find position and rotation of unorganized parts and hand the robot only what is pickable.
- Q2L / Q3L smart modules: Compact all-in-one AI cameras in metal housing — point at the scene, configure over Ethernet in a web UI. Camera-agnostic: USB3, GigE or even existing HDMI feeds.
- Robot integration: Mount QiCHECK on Wyzo, UR, FANUC, Doosan or Dobot arms for 360° inspection sweeps and vision-guided picking — one vision system handles the whole control loop.

## Case studies (in production)
- [Furniture components — Packaging & closed-box inspection for furniture components](https://www.aicobotix.com/case-studies/component-packaging): Every component package checked for completeness and content — including mirror-image L/R parts that human eyes struggle to keep apart. QiPACK on the conveyor, QiSHELL for sealed boxes. Key result: 98.9% live pass rate.
- [Power electronics — Robotic inspection of power-electronics assemblies](https://www.aicobotix.com/case-studies/power-electronics): A UR robot sweeps 360° around each power module — 75 checkpoints verified in under a minute. Key result: 75 checkpoints < 1 minute.
- [Enamelware — Surface inspection of high-gloss enamelled sinks](https://www.aicobotix.com/case-studies/enamel-surface): Mirror-gloss white enamel — the hardest surface for machine vision. QiCHECK grades every sink from both sides against per-surface quality rules. Key result: 9× 13 MPix cameras · defects < 0.5 mm.
- [Window manufacturing — Transparent glue-bead inspection for window bonding](https://www.aicobotix.com/case-studies/window-glue): A robot lays a transparent adhesive bead along the window frame — QiCHECK verifies application and distribution along the whole edge before the glass is pressed on. Key result: Tenths of a millimetre, full perimeter.
- [Component packaging — Vision-guided component packing at high line takt](https://www.aicobotix.com/case-studies/high-speed-packing): Four robot stations guided by QiCHECK place small components into packages at full line speed — while four Q2L smart modules watch conveyor fill so the line never starves. Key result: 4 + 4 vision modules on one line.

## More use cases (same platform)
- [Clean-room optical assembly](https://www.aicobotix.com/case-studies#optical-assembly) (0.1 mm guidance): A small SCARA robot with vision guidance assembles fine optical elements inside a clean box and checks quality in-process — batches of 100+ pieces without an operator.
- [Metal-clip check after robot insertion](https://www.aicobotix.com/case-studies#clip-check) (One-click teach-in): A Q2L smart camera on the robot arm verifies clip insertion in moulded parts from multiple angles. No programming — taught with pictures of good and bad parts.
- [Guided electronic assembly](https://www.aicobotix.com/case-studies#guided-assembly) (Every step verified): A Q3L module over the bench checks polarity, presence and part IDs at each assembly step, with a step-by-step operator screen and full archive.
- [Left/right pair validation](https://www.aicobotix.com/case-studies#lr-validation) (< 100 ms detection): Mirror-image L+R parts fool the eye, not the AI: orientation and count checked in under 100 ms, with an automatic line stop on a duplicate or missing part.
- [Stamping-line crack detection](https://www.aicobotix.com/case-studies#stamping-cracks) (At 0.5–1 m/s): A Q2L trigger camera plus high-resolution neural inspection catch openings and fine cracks on parts moving on the conveyor at full speed.
- [Rubber moulding defects](https://www.aicobotix.com/case-studies#rubber-defects) (Trained on NOK samples): Over-moulding, holes and cold moulding on black rubber parts — where good-sample logic is not enough, the system learns from defect samples too.
- [Bin picking of pouches](https://www.aicobotix.com/case-studies#pouch-picking) (80 ms response): An RGB-3D camera and QiPICK neural processing localize unorganized pouches for a delta robot — only the pickable top items are sent to the controller.
- [Label print verification](https://www.aicobotix.com/case-studies#label-verification) (Reference-based diff): QC sets one good label as reference; the system flags every misprint, missing symbol or wrong text across thousands of labels, marked in red.
- [Warehouse box counting](https://www.aicobotix.com/case-studies#box-counting) (Count in ~1 s): A high-resolution camera counts boxes inside packaging in about a second — inventory checks without opening anything.
- [Robot collision prevention](https://www.aicobotix.com/case-studies#collision-prevention) (Stops before damage): Tray reference targets are checked before a robot inserts material — deviations stop the robot and call maintenance via HMI or chat message.
- [Moulds on a carousel](https://www.aicobotix.com/case-studies#mould-carousel) (Multi-camera coverage): Material presence in automotive seat moulds — labels, springs, rubber — verified by multiple cameras across a rotating carousel.
- [Mobile QC station](https://www.aicobotix.com/case-studies#mobile-qc) (All-angle check): A semi-automatic mobile unit checks metal parts from all angles and sorts OK/NOK automatically — a universal end-of-line check.

## QiCHECK Assistant (MCP + LLM)
- [Ask anything. QiCHECK answers.](https://www.aicobotix.com/#assistant) QiCHECK-3 ships with an Analytics MCP server — the open protocol that connects AI assistants to live production data. Plug in Claude, ChatGPT or a fully on-premise LLM and ask in plain language. The answers come from your line, not from the model’s imagination.
- Computed, not guessed: Numbers come from deterministic queries over the same production archive your dashboard uses — the AI only explains and charts them. That is why the answers always match the dashboard.
- Every role, one question away: Plan burndown for the director, bottlenecks and line balance for production, defect Pareto with photos for quality, full serial-number trace for audits — and ad-hoc SQL for questions no report anticipated.
- Claude, ChatGPT — or your own LLM: The MCP server is client-independent: connect the assistant your company already trusts, or an on-premise LLM so production data never leaves the plant.
- Read-only, tokenized, audited: The service has no write operations. Per-user access tokens, every query logged, bound to your internal network — analytics, not control.
- No internal LLM? We train your own: Without an internal LLM and the tools around it, most of your production data stays out of reach. We deploy an on-premise model and train your own LLM agent — connected to your data over MCP, tuned to your processes and terminology.
- MCP tools include: production_pace, bottleneck_analysis, defect_analysis, quality_trend, trace_product, run_sql, + 12 more.

## Contact / demo
- [Book a 20-minute demo](https://www.aicobotix.com/demo): Twenty minutes, no prep needed on your side. We show QiCHECK running on real production and discuss your part, your line and your defects.