Today, a new company is emerging from stealth to help companies build open-source AI infrastructure and reduce engineering overhead, backed by Google's AI-focused venture fund.
Cake provides over 100 products for enterprises, including data source adapters (such as Apache Hadoop), data ingestion (such as Apache Kafka), data labeling (such as Label Studio), vector and graph databases (such as Milvus or Neo4j), and generative AI. Integrate and secure your components. APIs and related tools (such as Anthropic) and many other categories.
This suggests why Cake is called so. Cake takes the various “layers” that make up an AI stack and integrates them into a more understandable, production-ready format that is more business-friendly.
“Big picture issues”
Founded in New York in 2022 by Misha Hirske (CEO) and Skylar Thomas (CTO) (pictured above), Cake launched last year and already has companies including AI bioscience startup Altis Labs and data intelligence insurance technology Ping. We work with customers of However, the company has not made much noise in public so far.
In addition to today's official announcement, Cake announced that it has raised $13 million since its founding. This includes $3 million in pre-seed funding raised throughout its formative years and a recent $10 million seed round led by Google's Gradient Ventures.
“We're not being secretive. We're just building and working with customers,” Harsk explained in an interview with TechCrunch last week.
Herscu previously founded an AI company called McCoy Medical Technologies focused on machine learning infrastructure for radiology, which he sold to IT vendor TeraRecon in 2017. He then joined New York venture capital firm Primary Venture Partners as an “operator-in-residence,” talking to hundreds of data science and AI executives and driving their next venture.
“I made over 200 customer research calls and asked them what their biggest pain points and bottlenecks were,” says Harsk. “The biggest challenge wasn't a single part of the stack, like setting up a vector database or a data pipeline. It was that there were a ton of different components across a very rich ecosystem. How do we make sure everything is integrated and production-ready?”
This is what Harsk calls the “big picture problem” that his new business enters into.
Cake's purpose is to understand the myriad open source components that make up a modern AI stack and provide a bundled and managed open source AI infrastructure for small teams. This isn't about building a business around a single open source project, as countless companies have done. Instead, it's about assembling and delivering hand-picked open source projects across the stack and making sure they run smoothly.
A large financial services company has millions of documents containing complex financial data and runs RAG (Search Augmented Generation) on these files to improve the quality of responses to natural language queries. Suppose you want to. When off-the-shelf products don't do the job or are inappropriate for compliance reasons, companies must install and stitch together several different components to build their own system. This is a time-consuming task, but Cake can handle it.
Elsewhere, hospitals may need to build secure systems for analyzing images from CT scans, or e-commerce companies may need to upgrade their recommendation engines. These are all potential use cases for Cake.
“We work in all areas, but our sweet spot is when companies are going beyond what a simple off-the-shelf product can do,” said Harsk.
parallel development
Cake's CTO Thomas previously worked as chief architect at IBM and most recently as a high-performance engineer and strategy director at Hewlett Packard Enterprise, where he previously worked at MapR. I bought a company.
Having worked on hundreds of projects with customers large and small over the years, Thomas has noticed a trend that permeates nearly every project: Every company uses open source tools in some way. He realized that many of these results were new research results. Laboratory. Still, it hasn't been easy to use them in businesses.
“Even for the largest companies, it takes a tremendous amount of time to incorporate what comes out of the lab into their operations,” Thomas told TechCrunch. “A lot of that is because most of them aren't ready for businesses. There may be no certification or authorization, and businesses have to do it themselves.”
There are similarities to what Cake is aiming for here. Europe has companies like Finland's Aiven, a $2 billion unicorn, doing similar things with a focus on data infrastructure. Perhaps the most obvious comparison is Red Hat, which IBM acquired for $34 billion and is best known for its enterprise-grade Linux operating system (RHEL).
“In the early days of Linux, there were thousands of open source packages that everyone wanted to use, but they weren't integrated or secure,” Thomas says. “The Red Hats of the world made Linux secure for enterprises because there was no model to support it. We want to do the same thing for AI today.”
There are plans to eventually introduce a hosted version of Cake, but for now companies must run Cake in their own environments. For many people, this won't be a problem, as data privacy regulations mean they can't send the data outside their system anyway. However, the hosted version may be attractive to organizations with lower compliance obligations.
“It's actually easier if you can control the cloud,” Herscu added.
In addition to lead investor Gradient, Cake's seed round included pre-seed investors Primary Venture Partners, Alumni Ventures, Friends & Family Capital, Correlation Ventures, and Firestreak Ventures.
A previously unannounced $10 million seed round closed in April, demonstrating not only the founder's background but also the company's traction. Harsk said the company is already looking into its next round of funding, with tentative plans to raise again around mid-2025.
“From a traction standpoint, we're already close to being a Series A company. We were able to get there fairly quickly,” Harsk said. “If you go to Series A, it's probably going to look like Series B.”