Glimpse Takes Its Fast CT Software Beyond Batteries

Glimpse is launching Explore, a version of its CT image-processing software for quality checks outside batteries. TechCrunch

Explore is aimed at aerospace, automotive, electronics and medical-device companies. It works with CT scanners those companies already have in labs and development shops. Engineers use those scanners to build 3D X-ray volumes that show internal structures, welds, empty pockets and whether parts fit together, all without cutting hardware open.

The product is software, not new scanner hardware. Glimpse built its own processing pipeline, based on deep learning models, that compresses scans and removes noise. The first processing step runs on local edge computers at the customer site. That keeps very large scan files on site for the first pass and shortens the wait from scan to usable 3D image.

The method grew out of battery inspection. There, Glimpse increased throughput on existing CT scanners so battery makers could check cells 10 to 30 times faster. That lifted capacity from a handful of cells per day to tens of thousands per day. Throughput has been the limit. CT shows rich internal detail, but slow scanning and reconstruction meant most teams could only check samples.

Glimpse is also working with partners on what it calls a customer super scanner that could finish a scan in a second or two. That hardware is still in development. For now, Explore is for lab and development work, not for inspection directly on the production line.

Glimpse has more than 100 customers, including Anker, Lucid and the U.S. Navy, and revenue in the mid seven figures. Co-founder and CEO is Eric Moch. Alongside licensed software, the company runs its own CT scanner near Boston for its Scan On Demand service and is opening a second Scan On Demand site in the San Francisco Bay Area.

The broader context here is a move from sampling a few parts to scanning many more. Lab CT has long offered detailed views but taken too long for wider use. Teams scanned a few units closely, then depended on electrical tests, camera-based vision systems or teardown for the rest. If scan times fall by ten times or more, CT can support more design rounds, failure-analysis work and pilot builds without slowing schedules.

Looking at what this means for hardware teams, the local-processing design is as important as speed. It avoids sending terabytes of 3D data off site for every job and fits labs where a workstation sits next to an instrument. It also lets Glimpse improve the software without customers buying new scanners.

In my view, faster lab scans are useful but do not by themselves fix factory quality. Moving from tens of thousands of scans in a controlled lab to automatic pass-fail decisions on a factory floor brings extra work around part handling, traceability and false alarms. Explore will test whether one software pipeline can handle very different shapes and defects, from airframe brackets to medical implants. I have watched lab tools make this jump before, and when they do, they stop being specialist equipment and become everyday engineering tools.