Ceramic.ai by Anna Patterson: Boosting Enterprise AI Model Development Speed and Efficiency

Anna Patterson's journey through Silicon Valley is nothing short of impressive. She's the brains behind three startups, including the search engines Xift and Cuil, and recall.archive.org, which eventually morphed into the Internet Archive. She also held the position of vice president of engineering at Google before launching Gradient Ventures, an AI-focused seed fund. And guess what? She's still not done innovating.

In a chat with TechCrunch, Patterson shared that a few years back, she felt the urge to dive into something new but wasn't sure if she had another startup left in her. But then, after a breast cancer diagnosis in 2023 and spending much of that year away from work, she faced a choice: return to her old life or embark on a fresh adventure.

She chose the latter, and that's how Ceramic.ai came to be. This new venture aims to provide foundational AI training infrastructure that helps enterprises train large language models (LLMs) faster and with fewer GPUs than what's currently out there. Ceramic boasts that its model can handle long contexts, work with any clusters, and help scale models by a whopping 100 times.

The idea for Ceramic struck Patterson when she realized that the existing LLM infrastructure was too complex and variable for widespread enterprise use. She explained, "Whenever you’re trying to stretch an infrastructure 10x, it usually works no problem. But when you’re trying to stretch it, you know, 100x or more, you often have to kind of take a step back and rethink it. So, I thought to myself, if this is some infrastructure that we’re going to run for the next 10 years, is this how I would do it?"

That reflection led to the birth of Ceramic AI in January 2024, co-founded with Tom Costello, the company's chief scientist. They've been working under the radar since then and have already secured partnerships with AWS and Lambda, though they're not bringing in any revenue just yet. Patterson's strategy is to build credibility, awareness, and trust with potential customers before diving into sales.

Recently, Ceramic raised a $12 million seed round led by NEA, with contributions from IBM, Samsung Next, and Earthshot Ventures. Patterson chose NEA to lead the round due to their technical expertise. The funds will primarily go towards sales and further development.

So, what's different this time around for Patterson? After three startups, a career filled with senior roles, and years as an investor, she says her approach to building startups has evolved. She feels more time pressure now than when she was younger and building her first companies. "When I was younger building companies, I felt this, I mean this is kind of sad, but I felt this really relaxing time of like running a just engineering, no-sales company," she said. "Whereas now, you know after being a VC for a number of years, I definitely feel more time pressured."

But she sees this as a positive change, believing it's better to get your product in front of customers earlier, gather feedback, and iterate.

This approach is crucial for Ceramic, especially since it's not the only player in the game. Together AI, for instance, also aims to help companies "turbocharge" their model building and has raised over $530 million in venture capital. MosiacML, which focused on accelerating LLM building, raised $37 million before being snapped up by Databricks in 2023 for $1.3 billion.

Ceramic has its work cut out for it in this fast-paced market, but with Patterson's track record and vision, it's definitely a startup to watch.