Maximize Yield, Productivity and Quality using AI
SixSense’s AI suite turbo-charges fabs and OSAT lines by automating defect review at scale, tracing recurring faults to their root cause, and instantly flagging at-risk wafers. The result: less yield exposure, tighter quality control, higher productivity and higher confidence on every run.
Trusted by Top Semiconductor companies
Unlock 95% of your fab’s hidden intelligence
AI-ADC: End-to-End Visual Inspection, Automated.
AI-ADC is our end-to-end solution for automating visual inspection. With tools for training, deployment, and maintenance, it delivers faster accurate defect classification with fewer labeled samples—reducing cycle time and cleanroom footprint.
- Build AI is end-to-end environment for preparing training data with least effort and building high-accuracy AI models. It enables sampling of most useful images automatically and model training with minimal manual effort and exceptional accuracy.
- ClassifAI runs AI models in production from across AOI tools. It enables high-throughput and low-latency visual inspection with near-zero escapes.
- VerifAI is a smart toolkit to auto monitor and adapt production models — with drift detection and self improvement capabilities to ensure reliable performance in production
Build AI: Train smarter, not harder
Smart Data preparation
- Pick a diverse, balanced, de-duplicated set—including hard-to-find examples from millions of images.
- AI-assisted label correction at 10x speed.
- Launch once with 0% escapee in production; skip endless retrains and rollbacks.
Classification >> Detection and Anomaly detection
- True classification models—no patched detection or anomaly workarounds.
- Skip bounding boxes; data prep is quick and easy.
- High accuracy even on extremely small and similar looking defects
Large vision foundation models specialized for semiconductors
- Pre-trained on millions of wafer-defect images.
- Thinks like a PhD, not a high-schooler — learns faster, performs smarter.
- Delivers > 95% automation, 0% UR, and up to < 0.05% OR (using image count in just 2-digit instead of 4-digits per class).
Explainable AI
- Verify what your AI learns and launch it only when you believe it!.
- No black-box AI any more!.
- Look inside the model’s brainand see ‘what’ it recognizes as defect.
ClassifAI: Smart & Scalable Defect Classification — Built for High-Throughput Inspection Lines
Seamlessly run in production
- Single model covers hundreds of devices and tools across generations.
- Built-in support for multi-view, multi-tool and multiple magnifications.
- Ready integrations for Rescans, Klarf variants, and incoming defects—deployable by yield engineers.
Operator-First Production Interface
Seamlessly run in production
- AI auto-grades and funnels new defects to the built-in manual review interface - allows for 1-click retraining.
- Familiar Operator UI with minimal learning curve.
- Tools such as interactive wafer maps, hotkeys and color‑coded defect maps give a head start in adoption.
Lot Disposition & Root-Cause Analysis
Built on AI-ADC's trusted labels, Lot Disposition auto-decides pass, hold, rework, or scrap for each lot. It balances yield impact, recurrence patterns, and severity risk, then writes the verdict straight into your systems—eliminating manual sign-offs, avoiding costly excursions, and keeping cycle time lean even at peak throughput.
Dispose AI
Seamless lot disposition
- Surfaces critical repeating defects, wafer-map signatures, and yield outliers.
- Enables auto-lot checkout decisions for yield engineers based on yield criteria and low kill-ratio defects.
- Puts lots with low first pass yield or low final yield on hold for review.
CausifAI
Faster Root Cause Analysis
- Pin points key sources of failure by correlating process tool, lot history and inspection data.
- Quantifies wafer exposure via wafer-stack analytics.
- Reports and charts for yield engineers to trace back sources of true and false yield loss .
Predict AI: Predict and Pinpoint Yield Risks
Predict AI
- Forecast Defects Early – Predict die failure probabilities from inspection, metrology, and process data in early steps.
- Target High-Risk Areas – Identify layers, tools, and patterns driving repeat defects, and dynamically focus sampling where it matters.
- Prevent Scrap Before It Starts – Trigger proactive tool and recipe adjustments and reduce unplanned downtime with self-optimizing, data-driven production.
