SixSense LogoSixSense Text
March 2026 | NewsletterBy Prakriti ChaturvediLast updated: 27th Mar 2026

March was an exciting month for SixSense. We expanded our production deployments, strengthened our presence in the US and Taiwan semiconductor markets, and launched new technology for real-time inspection.
Key highlights included doubling inspection throughput and eliminating escapes for a global semiconductor leader in Asia, speaking at SEMI’s Smarter Sensors, Smarter Fabs in California, joining AppWorks Accelerator #32 in Taipei, and launching an AI system designed for embedded, high-speed inspection environments.
Together, these milestones reflect our focus: bringing production-grade AI into semiconductor manufacturing, where speed, accuracy, and reliability matter.

SixSense at SEMI: Smarter Sensors, Smarter Fabs

SixSense participated in SEMI’s Smarter Sensors, Smarter Fabs: AI at the Edge in Semiconductor Manufacturing event in Milpitas, California.
At the event, our founder Avni Agrawal shared her perspective on the growing relevance of AI-powered Auto Defect Classification for yield improvement. She spoke about the shift toward edge deployment of defect classification, with examples across wafer fabs, advanced packaging, and traditional packaging. It was also encouraging to see leaders from Onto, KLA, and Microtronic discuss the growing importance of real-time defect classification at the edge across optical, SEM, and other image-based inspection applications.
The event was also an opportunity for us to share our broader vision for inspection intelligence: using inspection signals to help teams make faster decisions. Over time, we believe this can move yield engineering beyond traditional SPC-based workflows toward more autonomous systems that detect patterns, link root causes, and drive earlier action.

SixSense Joins AppWorks Accelerator #32

SixSense has joined AppWorks Accelerator #32, a leading startup community headquartered in Taipei, Taiwan.
As SixSense continues to expand its presence in Taiwan, one of the world’s most important semiconductor hubs, AppWorks provides a strong platform for growth. It opens access to a broad network of founders, investors, and industry partners with deep roots in the Taiwanese market.
AppWorks is one of the most influential launchpads for tech and AI startups in Greater Southeast Asia, with over 400 alumni companies across its ecosystem. This gives SixSense stronger access to local market intelligence, potential enterprise customers, and strategic partnerships as we continue to scale in the region. We are excited to learn from the cohort, deepen our relationships across Taiwan and the broader ecosystem, and accelerate our growth in AI-driven yield intelligence.

SixSense joining AppWorks Accelerator #32. The design features a blue and white background with centered text reading “SixSense Joins AppWorks Accelerator #32.” Below, two rounded white cards display the AppWorks and SixSense logos side by side with an “x” between them, symbolizing the partnership. A green banner at the bottom reads, “Scaling AI innovation in semiconductor manufacturing globally."

Hierarchical AI for Real-Time Wafer Inspection

This month, we launched a high-throughput AI system designed to operate directly inside inspection tools under strict production constraints for speed, accuracy, and reliability.
The system achieved sub-20 millisecond decision times per image, 98.2%+ classification accuracy, zero escapes, and reliable deployment on embedded hardware.
To meet these requirements, we developed a hierarchical AI architecture. It starts with a semiconductor-trained foundation model and distills that knowledge into a smaller, production-ready model to reduce compute cost and footprint. The system is then fine-tuned for process-specific contexts and uses dynamic routing, so most images take lightweight paths while complex cases receive full compute. This enables high accuracy and machine-speed decisions in production environments.
This launch marks an important step toward making real-time, in-tool AI practical at scale for semiconductor inspection.

Infographic titled “Constraints of Real-Time Inspection” illustrating the challenges of semiconductor wafer inspection systems. On the left, a high-speed inspection tool scans a semiconductor wafer on a 400mm inspection stage, generating a continuous stream of high-resolution image data. In the center, the image data flows into an embedded computer module with limited CPU cores, restricted RAM, and no GPU support. On the right, the system runs an optimized algorithm and decision-tree classification process to determine whether wafers are defective or good. The graphic highlights a requirement of processing each image in under 20 milliseconds while maintaining 98.2% accuracy.

Read full blog:
https://www.sixsense.ai/blog/foundation-model-to-production-speed-inference-in-the-fab/

Doubling Throughput and Eliminating Escapes for a Global Semiconductor Leader

Through a close partnership with a leading AOI OEM in Penang, one of the largest providers of post-wire-bond inspection systems, SixSense deployed an edge AI solution for a global analog and embedded semiconductor leader.
The customer faced a serious quality risk. Critical defects such as depressed wires were escaping from the wire bond step, raising the risk of major quality escalations. At the same time, the existing inspection system had low throughput and depended heavily on operators for AOI image review.
By integrating AI-based defect classification directly inside the OEM machine, SixSense doubled inspection throughput, eliminated defect escapes in validation, and removed dependency on manual review.
Delivered in under 3 weeks, this deployment shows the value of close collaboration between AI software providers and equipment OEMs: faster deployment, tighter integration, and immediate production impact.

this deployment shows the value of close collaboration between AI software providers and equipment OEMs: faster deployment, tighter integration, and immediate production impact.