Originally published in Petacrunch on August 9, 2019.
PetaCrunch: How would you describe Latent AI in a single tweet?
Jags Kandasamy: Latent AI brings Adaptive AI™ to the edge through our core technologies and platform tools enabling efficient, adaptive AI optimized for compute, energy, and memory with seamless integration to existing AI/ML infrastructure and frameworks.
PC: How did it all start and why?
JK: My previous startup, OtoSense, acquired by Analog Devices in early 2018, focused on automated human hearing. At OtoSense, we truly understood the need for edge inference. Post exit, I joined SRI International as an Entrepreneur in Residence (EIR). SRI is the birthplace of Nuance Communications (NASDAQ: NUAN) and Siri (acquired by Apple). I was immediately drawn to an SRI-developed technology that was a DARPA funded initiative built for efficient edge AI inference. The next generation of computing is going to be built on AI and it was clear we had the opportunity to make a significant impact with tools that democratize AI development.
PC: What have you achieved so far?
JK: In less than a year, we assembled an incredibly strong core team. My co-founder and CTO, Sek Chai, came from SRI as the lead Program Investigator for DARPA related projects. We recently opened our engineering center in Princeton, New Jersey to take advantage of the east coast talent. We came out of stealth in June and just announced our first product, LEIP (Latent AI Efficient Inference Platform) Compress – Post Training and Training Aware. LEIP Compress allows our customers to compress their trained DL models with proven results of 6-10x compression and very little accuracy loss. LEIP is an integration of AI training and a compiler combined into an automated framework that generates optimized code based on a deep neural network model, parameters and available hardware resources. We are actively working with several Fortune 500 customers around the world to pilot our technology.
PC: How will you use your recent funding round?
JK: Our goal with this funding is to identify the product-market fit for our tools and successfully execute a few pilots. We are also developing additional products, LEIP Compile and LEIP Adapt, and details will be coming out in the next few months.
PC: What do you plan to achieve in the next 2-3 years?
JK: Our plan is to first identify our product-market fit and then start seeding that market. Concurrently, we will focus on growing our partner ecosystem. Our vision is that Latent AI becomes synonymous with Edge AI. We intend to unlock the power of edge AI, enabling industrial IoT, smart cameras, voice assistants for smart homes and autonomous vehicles, among other use cases.