Beyond Buzzwords: Glasswing AI palette guides startups navigating AI’s diverse terrain

For the past year, hype around generative AI has reached a fever pitch, as large language models (LLMs) have dominated the news cycle.  According to Glasswing Ventures founder and managing partner Rudina Seseri, while valuable, generative AI when used as a shiny wrapper will not be defensible on which to build a business.

Today the early-stage venture capital firm released its Glasswing AI Palette as a new framework to help educate and inform startup founders about the broad set of technologies that modern AI and machine learning (ML) encompasses and how those tools can solve real enterprise challenges. 

Glasswing is no stranger to the AI market, with the firm raising $112 million in 2018 to fund AI startups in an era long before anyone had ever heard the name ChatGPT. In 2022, the firm raised an additional $158 million to further its mission of helping to build and grow data and AI technologies. Among the many companies that Glasswing was an early investor in are data technology vendor ChaosSearch, feature engineering company FeatureByte and enterprise AI data vendor Causely.

“Everyone is excited about gen AI and don’t get me wrong, we are too,” Seseri told VentureBeat. “But there’s a lot that’s happening with other techniques.”

What the Glasswing AI Palette is all about

The Glasswing AI Palette aims to provide a common language and framework to help navigate the complex and ever-evolving landscape of AI.

Seseri explained that the palette maps AI techniques like transformers, Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) to corresponding data types and potential use cases. The tool also links out to Glasswing’s ongoing ‘AI Atlas’ project, providing more in-depth explanations and evaluations of different techniques and how they can be used in applications.

While much of the hype in the industry focuses on generative AI and LLMs, Seseri is eager to emphasize the breadth of AI beyond these buzzwords. As an investor, she emphasized that Glasswing is focused on companies delivering value for the enterprise and security markets.

“AI and gen AI are not necessarily the same thing,” she said. “Startups using the framework may find opportunities beyond API wrappers in applications of techniques like CNNs, RNNs and transformer combinations.

Image: credit: Glasswing Ventures

The problem with LLM wrappers? They’re indefensible

A far too common pattern with startups today is taking an existing LLM and creating a wrapper around it.

Seseri said what the ChatGPT era has enabled is the mass consumerization of AI. While that can offer productivity benefits for organizations, in her view using an LLM wrapper approach isn’t defensible from a business perspective and organizations won’t be able to build some form of competitive moat that will enable differentiation.

The types of companies that Glasswing is looking for and hoping to enable are those that are using algorithms, techniques and architectures at the core of the product to provide an output that otherwise wouldn’t be achievable. That output should be something that solves a big problem, whether it’s on the enterprise or the security side with big budgets in a ‘must have’ category.

“What we look for is truly AI native companies, where they’re solving problems that otherwise would not be able to be solved,” she said.