The Latent AI claims it can reduce common AI models. Lands some important backing, 10x

Latent AI was a Menlo Park-based startup that started in 2003. It pitched investors during TechCrunch’s Battlefield contest. It didn’t win that contest, but that hasn’t kept it from winning the interest of investors elsewhere. The company just closed $19 million of Series A financing in a round led by Future Ventures. It also had participation from 40 North Ventures (Booz Allen), Lockheed Martin and Autotech Ventures. It has raised $22.5million.

How much are these backers actually funding? According to the company, its software tools can run AI models anywhere regardless of hardware limitations, even on inexpensive chips found in edge devices. The company claims it can reduce common AI models up to ten times with no noticeable decrease in accuracy. This is partly due to an “attention mechanism”, which allows it save energy and only run what it needs, and partly due its ability adjust its workload according the operational context and environment.

According to Latent AI, this means that the tools are able to help developers create AI models optimized for computation; they can overcome power and memory constraints; and there is almost no latency (hence the company’s name).

Steve Jurvetson is a veteran investor who cofounded Future Ventures. He says that face detection algorithms can be run locally in security cameras, appliances or voice interfaces like Siri, and they work instantly even if there is no internet connectivity.

There is a large market for Latent AI’s tech. Although Jags Kandasamy, the cofounder and CEO of Latent AI, declined to discuss specifics in an interview, he indicated that the U.S. government was already a client, in part due to strategic investors such as Booz Allen. In a press release about Booz’s investment into Latent AI last month, Steve Escaravage (a senior vice president at the government service agency) stated that “the ability to quickly collect, analyze, and act upon data is the heart” of America’s national defense strategy.

Lockheed Martin is another strategic investor. He’s very interested in finding ways AI technology can improve the U.S. Military’s situational awareness on land, water, space and cyber domains. It’s not hard to see why it would be attractive to the startup.

Kandasamy spoke this afternoon also about an unknown ski manufacturer who is using Latent AI tech in its Google Glass-like Augmented Reality Goggles. He suggested that Latent AI can see a lot of potential in the commercial space, as well.

Of course, given that the world is now rife with data-collecting-devices and that there’s an enormous interest in making that data actionable without having to send it back and forth to a distant cloud, there are many other companies and projects with the same objective as Latent AI. Open source tools such as TensorFlow and hardware sellers like Xilinx are just a few of the many.

Kandasamy believes that everyone is not perfect. TensorFlow’s Kandasamy says developers only have community support for production deployment. The chipmakers around the globe are focused only on the hardware they make and have not integrated vertically. What about their rivals? Kandasamy says that they are either focused on compiling or compression and not both.

Regardless of the outcome, Latent AI is a company with a rich backstory. After Kandasamy, a serial entrepreneur, sold his last startup to Analog Devices, he headed to SRI International in 2018 as an entrepreneur-in-residence. Sek Chai was the technical director of SRI International for almost a decade. He specialized in machine learning, low-power computing and computer vision. Soon after Kandasamy’s death, Chai formed Latent AI, and began to iron out the details.

With new funding, and a growing client base who pay Latent AI monthly to have access to the tools and then deploy them on-premise, the question now is whether or not the company of 15 people is moving fast enough to keep up with its future and current rivals.

Jurvetson clearly believes it is well-positioned. Jurvetson says that he is an SRI International advisory board member since more than 10 years. He has seen many technology like Siri evolve and later spin off of the company.

He says, “This is my only investment in”

Publiated at Tue 10 August 2021, 04:04:05 +0000

Leave a Reply

Your email address will not be published. Required fields are marked *