3D parts-scanning technology slashes production line downtime
5 March 2025 
A Californian start-up, called Backflip AI, has developed a technology that it claims will cut production line downtime dramatically by allowing anyone to create models of replacement parts automatically from 3D scans, which can then be manufactured rapidly using techniques such as CNC machining or 3D printing.
The technology will help manufacturers to compress the time from a part breaking to creating a digital 3D model of that part, helping them to get production back online faster.
Assembly lines are composed of tens of thousands of unique parts and it is impossible to stock spares for all of them.
To complicate matters, many production lines are built by third-party systems integrators who do not provide CAD models to their customers. Some components are no longer made, and machinery often pre-dates digital designs. As a result, manufacturing teams spend thousands of hours manually recreating 3D models for parts that are already in front of them.
Backflip says that its new tools will eliminate this headache, automatically converting 3D scans of parts into manufacturable CAD files in about a minute.
At present, if maintenance teams do not have a replacement part, they will often have to build one from scratch, requiring someone to:
- measure the part and design a digital 3D CAD model;
- manufacture the part at an internal or external machine shop; and
- install the replacement and restart the production line.
Each of these steps can take hours. Unexpected downtime can cost $3m an hour and manufacturers suffer more than $50bn in losses every year because of it. Backflip says its new tools can slash the first step to minutes, cutting downtime by as much as half, and bringing operations back online much faster.
The company has announced two tools powered by its AI model:
- a Solidworks plug-in that translates 3D scan data into parts files; and
- a Web app that converts 3D scans into parametric CAD files.
Both tools take what was once a complicated, costly process and collapse it into a simple operation that takes seconds.
The Solidworks plug-in shows each step in the process to build a part’s geometry and generates a native feature history that users can modify. This allows CAD designers to fine-tune the generated 3D model.
For new users, walking through the design process helps them to understand how the part was created, easing the learning curve for 3D designing and bringing more people into 3D CAD programs such as Solidworks.
Backflip says its development will reduce the technical barriers for designers, engineers, and technicians to recreate 3D models of parts that they don’t have CAD files for. The tools also make 3D scanning much easier.
“3D scanners map the surface of an object with incredible precision, quickly generating millions of data points, but they produce micro surface textures that can’t be manufactured with traditional tools,” says Backflip’s co-founder and CEO, Greg Mark. “Our technology automatically converts these intricate surfaces into clean geometries designed for existing 3D CAD and manufacturing software.”
The AI model was trained on a synthetic 3D dataset of more than 100 million 3D geometries – said to be the world’s largest.
“These new developments are the result of Backflip’s R&D investments in neural representation of 3D data,” explains the company’s head of AI, Logan Ford. “We’ve paired state-of-the-art AI/ML techniques with the team’s deep background in manufacturing to create a really powerful technology foundation. Our additional focus on resource-efficiency enables us to advance rapidly, and yields 60x more efficient training, 10x faster inference and 100x the spatial resolution of existing leading methods.”
Backflip recently emerged from two years of stealth development. It has already raised more than $30m of funding. The team behind Backflip previously invented 3D-printed carbon fibre, founded a business called Markforged, and took it public for $2.1bn.
