This story dates back to nearly 12 years ago, when Tayo Adesanya began working in the field of microchips and AI processors. His main role was to assist large manufacturers in deciding which chips to use in their hardware. Adesanya told TechCrunch that those years allowed him to foresee the trends in the AI computing market demand at an early stage. “Founding Lola Vision Systems was a bet on where the world was going to go and based on everything I observed,” he said.

In 2024, he launched Lola Vision Systems, which is a AI infrastructure company that develops software and chips for running AI models on devices. Its core product is a software that can convert AI models into instructions that can be executed by specific chips. Adesanya refers to this software as a "compilation toolchain" and notes that it represents a significant bottleneck: manually configuring a AI model for new hardware can take "about 200 hours" just to begin testing. Lola Vision indicates that the company has rebuilt this software layer and is also developing its own semiconductor chips, with the goal of automating more processes. Customers provide their own code and the AI models they wish to use, whether they are self-developed or open-source, and the software will convert both into instructions that can be executed by the customer's chips.

“Speed is just one part of it,” said Adesanya. He explained that a faster deployment speed allows aerospace companies and “other mission-critical organizations” to have the time to “run more accurate models on their own data, with lower power consumption.”

"For these customers," he said, "accuracy and reliability are not just icing on the cake. They determine whether a product can pass regulatory reviews and whether it can operate reliably in real-world environments."

Based in Washington, D.C., Lola Vision is one of several startups that attempt to provide NVIDIA technical alternatives for running AI on devices. Adesanya states that currently, many companies start with Nvidia's Jetson, which is a series of compact computing modules designed to run AI on devices, or they begin with open-source AI models. Adesanya mentions that these solutions "often have issues right out of the box or perform very poorly, so teams spend days or even weeks just to get them running, and then additional weeks debugging until the models are usable."

"Even so," he continued, "power consumption often exceeds the budget for edge computing, or the circuit boards are not capable of providing the computational power required for medium to large models to run successfully. This can result in recognition models falling behind their targets or misidentifying objects." (Edge computing refers to running AI directly on devices, such as cameras or drones, rather than in remote data centers. Recognition models are AI systems used for identifying objects.)

According to the company, 12 corporate clients have signed letters of intent, stating that they will make purchases once the chips from Lola Vision are launched on the market. The company currently has one signed client as well. The company has also established a partnership with SCALE – a microelectronics workforce development project – in order to collaborate with more semiconductor laboratories. “In order to generate revenue more quickly, we will now license our software on existing hardware,” said Adesanya. (In other words, the company will not wait for its own chips to be launched; instead, it will allow clients to pay to use its software on chips that already exist.) He added that the company has raised a total of just over $1 million in funding so far.

Lola Vision has been selected for this year's TechCrunch Battlefield 200, a project consisting of 200 companies chosen from numerous startups. "Even when I was still a student at Purdue University, TechCrunch was already one of my favorite media platforms," he said. After about a year of product development and signing the first customer, he felt it was time to apply for Battlefield to expose the company to a wider audience.

As for what he looks forward to the most about this event, he said it is “to establish meaningful connections and to learn as much as possible about what is happening both within and outside our field.” “And, to be honest, I’m also looking forward to investors writing checks,” he added.

If you want to learn more about Lola Vision Systems and the dozens of carefully selected startups (along with venture capitalists who will be visiting for inspections) that will join us in TechCrunch Disrupt next week, we welcome you to attend the event in San Francisco from October 13th to 15th.