ACCEL-KS Profile: MLNavigator

This profile is part of our series highlighting the recipients of the ACCEL-KS Proof-of-Concept grant, a partnership between the Kansas Department of Commerce and Groover Labs. ACCEL-KS provides up to $25,000 in grant funding to help Kansas innovators move their ideas from concept to commercialization. Over the coming weeks, we’ll introduce you to each team and the problems they’re solving.

Up next? MLNavigator, Cof-Founders James Auchterlonie and Donella Cohen

Co-Founders of MLNavigator, Donella Cohen and James Auchterlonie

Company name? MLNavigator

Product name: adapterOS

What gave you the idea for your product or startup?

Working in aerospace engineering, James began crafting an AI solution for workflow inefficiencies he observed on the job. He described the concept to Donella, who had been working at Microsoft during Copilot rollout and was excited about applying her product experience to the business. In researching the idea, top themes emerged around security, compliance, efficiency, and responsibility dependencies, leading to the development of an offline, private AI solution — adapterOS. 
Targeting small to medium businesses, the solution would have to be compute efficient, which further led James to tackle the token efficiency problem. Addressing security concerns, James engineered an AI orchestration system that is deterministic and yields a private, efficient AI inference engine that is mathematically provable, luggable, and reproducible. What has emerged is an orchestration system that enables the efficient creation and management of adapters to fine tune AI models.

What makes your product stand out?

AI prompting has been expanding in an effort to find real productivity benefits in practice. And prompting is expensive. Our solution, adapterOS, enables the orchestration of adapters to change the math that determines responses from LLMs. The unique architecture of this AI operating system creates an immutable record of the interaction, while also recording responses for future re-use and learning, resulting in significant reduction in token usage. While our user interface is heavily geared toward AI engineers, we are developing a lighter user interface for other target personas.

What have you learned so far on your journey?

The business community is hungry for compliant, private, reliable and cost-effective AI solutions that allow them to interact with their business data in meaningful ways -- not just generating emails and slide decks. Energy consumption is a growing concern as AI adoption expands. It’s important to us to harness the true power of AI as a tool, while making it sustainable. The possibilities for application of this technology are nearly endless. Zeroing in on the best markets and persona targets becomes more challenging as adapterOS’s potential grows.

How are the ACCEL-KS grant funds helping you reach the next stage of development?

Rapid acceleration of our product development is directly due to the ACCEL-KS grant funds enabling us to purchase software and services, hardware and office equipment. The grant also will enable us to file a provisional patent, protecting our IP and potentially providing additional commercialization or investment opportunities. The majority of funding spent to date has been for access to high-powered AI models to speed up coding, testing, and de-bugging.

What kind of support do you need from the startup community?

At this point, we are looking for expertise in several areas, including manufacturing, medical office management, as well as government security compliance, systems testing, and analytics. Interviews with these experts will help us better understand the landscape. We also are seeking additional funding to procure hardware and services vital to the continued development of adapterOS.

What is something interesting we should know about you or your project?

We want to change the way the world uses AI. By building a platform that makes specialized AI training and meaningful interactions with data more accessible, MLNavigator hopes to lead the next wave of machine learning innovation. We are deeply conscious of the potential energy crisis in the AI landscape, the specter of harmful AI and the largely ungoverned explosion of this technology. Our guiding questions as we develop this platform are “How can we help make AI safer, more reliable, more sustainable, more productive, and more accessible?”