Multimodal defect vision
Learned SPI/AOI/AXI/coating detection and classification.
AI inspection that sees and predicts solder defects. AOI, X-ray BGA voids, and coating coverage, without the false-call tax that floods engineers and hides real escapes.
Inspection is where factories bleed. Rule-based AOI/AXI floods engineers with 5–20x false calls while genuine escapes reach the field, and static thresholds never transfer across products, packages, or lighting.
Voidex ingests SPI/AOI/AXI and coating imagery, predicts defects with learned models that cut false calls while catching escapes, auto-generates inspection recipes, and root-causes every call back to print, place, and reflow.
Learned SPI/AOI/AXI/coating detection and classification.
X-ray head-in-pillow and cold-joint analysis.
Confidence scoring with human-in-the-loop routing.
New-board recipes generated from CAD and paste data.
Defects fed back to Stencix, Reflux, and Yieldix.
Conformal coating thickness sensing with cited write-back.
It shares perception, twin validation, and orchestration with the rest of Circuex, so every action is grounded and every result feeds the next board.
You choose how much control Voidex takes, and it earns the next level by proving itself.
Voidex perceives and predicts with zero write-back. You validate its calls against reality.
It surfaces grounded recommendations with the evidence behind each one.
Operators approve actions one at a time under policy, building trust and data.
Voidex tunes autonomously within guardrails, escalating only exceptions.
Trained on DGX/HGX, served at the edge on Jetson and IGX, validated in Omniverse and Isaac Sim.
Once Voidex went closed-loop, the stage stopped being the bottleneck. It just holds yield, and tells us why when it doesn't.
Book a working session and watch Voidex perceive, predict, and close the loop against your line data.