Larger parts, wider variance
Bigger, less uniform batches make consistent surface and dimension checks hard to hold by hand this is where inline inspection earns its keep most at scale.
Casting QC
Catch the defect before it leaves the line.
AI-enabled inspection for surface, dimension, and orientation hardware included.
Trusted by Global Enterprises
What we inspect
Five checks. One inspection pass.
Every casting is checked for surface defects, dimensional accuracy, orientation and alignment, part marking, and feature presence in a single pass, on every part.
Surface Defects
Cracks, porosity, inclusions, and flash flagged as the part comes off the line.
Across Casting Processes
Different process, different failure points
The three checks apply everywhere but what tends to go wrong shifts depending on how the part was cast.
Fast cycles, no time to spare
High-speed, high-volume production leaves little room for manual checks between cycles. Inspection runs inline, matched to cycle time, catching flash and surface imperfections without slowing the line.
Tight tolerances, fine detail
Complex geometry and tight tolerances mean small surface and dimensional deviations matter more, since many parts skip further machining entirely.
Deployment
Comes with the hardware. Fits your existing line.
Scanflow isn’t just software camera and sensor setup is part of deployment, fitted onto the line you already run.
Hardware setup included
Cameras, lighting, and edge devices installed and configured as part of deployment. No separate procurement or integration project on your end.
Fits your existing line
Mounts onto the conveyors and inspection points you already have installed without stopping production.
Edge, cloud, or hybrid
Run inference on-site for air-gapped lines, in the cloud for multi-site visibility, or a mix of both your choice.
Connects to ERP / MES / PLC
Inspection results flow into the systems you already run real-time pass/reject signals, no manual re-entry.
What else comes with it
Two more tools that come with it
Inspection is the core these make the data useful day to day, without adding headcount.
QC Agent
Ask your inspection data questions in plain language defect trends by shift, pattern, or time period without building a dashboard query each time. A natural-language interface over your stored results, not an autonomous decision-maker.
Data Annotation Tool
Label and organize defect images to calibrate detection models as new patterns or defect types come online built for your quality team to use directly, no data science background required.
How it works
From the line to a logged result
-
Step 01
Capture
Camera or mobile device captures every part as it passes.
-
Step 02
AI Inference
Runs on-device or on a local workstation no cloud round-trip needed.
-
Step 03
Detection
Defect identified in real time surface, dimension, orientation, OCR, or presence.
-
Step 04
Rejection
Auto-reject triggered pass/fail signal sent straight to the line.
-
Step 05
Dashboard
Results stored and synced audit-ready, searchable, with drift alerts by pattern.
Why it matters
What changes when inspection moves to AI
Reported industry patterns when casting lines add AI-enabled visual inspection directional benchmarks, not Scanflow-specific claims.
| What changes | Manual visual inspection | AI-enabled inspection |
|---|---|---|
| Coverage | Sampling — a fraction of parts checked | Every part, every shift |
| Consistency | Varies with fatigue, lighting, shift | Same standard, every time |
| Surface defect detection | ~40–55% on complex castings | Meaningfully higher, calibrated per part geometry |
| Cycle time impact | A separate manual checking step | Runs inline, matched to line speed |
| Documentation | Manual, batch-level logs | Automatic, part-level, audit-ready |
Scrap rate reduction reported on AI-inspected casting lines
Yield improvement reported on modernized lines
Of parts get a logged result, not a sampled subset
Who it's for
One inspection standard, across industries
The same checks apply whether the part is headed to an automotive assembly line, heavy industrial machinery, or high-pressure fluid-handling systems.
Frequently Asked Questions
Does Scanflow catch internal defects like shrinkage or internal porosity?
No internal and sub-surface defects need NDT methods such as X-ray/CT, ultrasonic, or eddy-current testing. Scanflow covers the surface, dimensional, and orientation layer, and is built to sit alongside your existing NDT process.
What casting processes does this work for?
Die casting, sand casting, investment casting, and gravity or permanent mold. Inspection points can sit post-cast, post-fettling, or pre-machining.
Do we need to source and install the cameras ourselves?
No hardware setup is part of deployment. We install and configure the cameras, lighting, and edge devices on your line.
How long does deployment take?
Around 4 weeks from kickoff to live inspection, including setup and model configuration for your part geometry.
Does it integrate with our existing ERP, MES, or PLC?
Yes inspection results and pass/reject signals connect directly to the systems you already run, without a manual re-entry step.
