[ad_1]
As large companies grapple with how to integrate AI into their platforms and operations, they have run into a problem: generative AI needs memory and its training data must be constantly updated in order to have any practical use. This area is now called “Live AI” and a number of startups are working in this area, including cohere and author. last, Trackjust raised a $10 million seed round to build living AI systems that, the company claims, think and learn in real time like humans do.
The round was led by TQ Ventures, with participation from Kadmos, Innovo, Market One Capital, Id4, and angel investors. Another investor in Pathway includes Lukasz Kaiser, co-author of transformers And the lead researcher behind OpenAI’s GPT o1.
Pathway’s offering includes what it calls “infrastructure components” that power live AI systems, feeding off structured and unstructured data, meaning enterprise AI platforms can make decisions based on up-to-date knowledge. Clients so far include NATO and the French post office La Poste.
“The way deep learning and LLMs work is that you take the training data and then you train the models,” Zuzana Stamirovska, co-founder and CEO of Pathway, told TechCrunch over a phone call. “But the question is, how do we deal with knowledge, how do we deal with memory? The current LLM holder acts like a very smart intern on the first day of his job, as he is offered a book to read but they can’t really memorize it and plus it’s not live, it’s static.
To address this, she said, the Pathway “enables developers to build a path through which they can feed live data into AI systems. We now do this during the prompting phase when building LLM or Gen AI applications.
Stamirovska – who is relocating to Menlo Park, California – has assembled an impressive, high-tech team to achieve the startup’s goals. Its co-founders are CSO Adrian Kosowski and CTO Jan Chorowski, who previously worked with recent Nobel Prize winner in physics and “Godfather of AI,” Geoff Hinton. Stamirovska herself is the author of a sophisticated prediction model of the complex network that existed in maritime trade, which was published by the US Academy of Sciences.
“The company started with an idea that came to my mind one sunny morning in Chicago,” she said. “I was there with a friend to a theoretical computer science conference… We had a little disagreement, and I said I should start my own business. So, I got out my laptop and started writing to people in my network about how to move forward with this “I still remember the taste of coffee at that moment.”
I asked her where she sees Pathway compared to other startups in this space? “For use cases in GenAI engineering and knowledge management, Cohere and Writer appear alongside us in the latest Gartner Quadrants,” she said. “While in enterprise deals, we often come across Palantir for AI transformation bids, even though they are less product-oriented than us.”
Commenting on this, Schuster Tanger, co-managing partner and co-founder at TQ Ventures, said: “Zuzana and the team at Pathway have cutting-edge insights and expertise in one of the most exciting areas in modern business… Finally, just barely. At the very least, the response from the developer community has been strong.
[ad_2]