dimensionalm
DimensionalMechanics co-founders Rajeev Dutt and Dave Hebert

DimensionalMechanics, a Seattle-based artificial intelligence startup that just emerged from stealth mode in March, today released beta software aimed at specific groups of customers — broadcasters, the entertainment industry and the fashion industry.

Examples include algorithms that can detect nudity in images, analyze the flow and translucency of fabric, and predict which variations of a news headline will attract more readers.

Co-founders Rajeev Dutt and Dave Hebert characterized the move as step toward making their products generally available through a software-development kit in the fourth quarter.

“We’ve pointed our arrows at a couple of different markets,” said Hebert. Dutt added, “This is the first time we can say we have a path to production for those companies.”

The company is releasing three products:

NeoPulse Profile identifies and classifies unique elements within data. For example, after analyzing 30,000 digitized text news articles and user patterns in reading them, it was able to predict which headlines would attract the most clicks. It can identify nudity and other inappropriate content in images and will soon be able to do so in videos. It can identify dogs by breed in videos, predict how popular an image will be on Instagram and reconstruct degraded images. Profile could also be used to filter out unnecessary data and detect patterns in military or patient data, the execs said.

dimensionalmechanics11NeoPulse Expert explains why certain patterns emerge within data, and recommends changes. For example, it can recommend edits to headlines to increase readership, a quality it gained after studying identical news stories bearing different headlines. It learned that a story headlined “Plane crash in Turkey kills 2” drew fewer readers than one bearing the headline “Plane crashes in Turkey,” because the latter headline drew in readers curious to know whether the crash had been fatal. NeoPulse Expert could use Form 10-K filings to predict why a company might succeed or help a physician determine why a surgical procedure is likely to be effective, the company said.

NeoPulse 3D simulates the behavior of objects and materials. For example, it understands how cotton, silk and other fabrics drape, flow and billow, and it knows how translucent they are. This could help a fashion company model new garments without having to actually construct them.

DimensionalMechanics’ offerings are platform as a service (PaaS). The company gives customers the option of running a server on-premises that pre-processes and stores the data, eliminating the need to hand over confidential information that is common in AI applications, the execs said.

The company has received $4.7 million in financing to date and currently employs 11 people.

Like what you're reading? Subscribe to GeekWire's free newsletters to catch every headline

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.