On Tuesday, Santa Monica-based Salt AI announced a research partnership with the Ellison Medical Institute in Sawtelle to develop a drug discovery platform.

That sentence is a breath of fresh air for one of the company’s cofounders, who once had a hard time describing Salt AI to others.

“If you say you’ve got a platform that does drug discovery and concert visuals, it’s kind of hard to wrap your head around for an investor,” Aber Whitcomb said.

But the fact of the matter is, Salt AI can do a lot – early iterations created non-playable characters for mobile games, an artificial intelligence-abled social network and, yes, concert visuals. Salt is an AI workflow platform that allows game designers, scientists, researchers – basically anyone that isn’t an engineer – to collaborate on custom AI solutions and workflows using many large language models.

“It brings AI development to people that would otherwise wouldn’t really have access to it,” Whitcomb said.

History of Salt AI

Whitcomb, perhaps best known as a cofounder at Myspace, started Salt AI with fellow Myspace veteran Jim Benedetto in 2023. The concept for Salt AI was born out of Whitcomb’s Jam City, where the company was experimenting with AI in game development.

“One of the challenges that we had was people didn’t feel included in the development environment soon enough with (artificial intelligence). And AI is somewhat of a black box, so it can be threatening to people – like, wait a minute, that’s my job, or I don’t know how that works,” Whitcomb said. “I found that getting more team members involved in the development of AI early gives a better chance of success.”

A peek under the hood

The flexible nature of Salt AI led to the partnership with the Ellison Medical Institute, which allowed biologists, chemists and researchers to tinker with the same mysterious algorithms once reserved solely for engineers.

AI-based drug discovery platforms have flooded the market in recent years. In 2024, most of the AI activity in the health industry went toward biotech, which received nearly half of the $10.4 billion venture dollars from that year, according to PitchBook. AI drug discovery saw its second-best funding year ever in 2024.

“Especially in AI drug discovery, people treat it like a big black box and a lot of them advertise they have a better model,” said Nate Beyor, head of life sciences at Salt AI. “And I’d see a lot of investors in particular look at it and they don’t really know how to make sense of it. It’s like everyone’s claiming they have a better mousetrap.”

Salt AI’s method allows the creation of different models that work together in drug discovery – the input data, a diffusion model, a sequencing model, a validation model and a docking model.

“This is an example of these Lego pieces kind of fitting together to solve this one particular problem,” Whitcomb said. “But the beauty of (Salt AI) is it can theoretically solve infinite problems, right? If you have your own proprietary model, you can import that as its own node in Salt and apply that to your use case.”