CreatorML

Overview

CreatorML was a New York-based B2B SaaS company founded in 2022 by Charles Weill, a former Google Research machine learning lead, that used ML models to help YouTube creators predict how many views a title or thumbnail change would generate before publishing.Accepted into Y Combinator's Winter 2023 batch and backed by ~$500K from YC, Amino Capital, Goodwater Capital, and Pioneer Fund, the company built a technically sophisticated product suite targeting YouTubers with over 100,000 subscribers.

Despite landing credible enterprise customers and generating what appears to be meaningful early revenue, CreatorML was listed as inactive on YC's platform by 2024, with founder Charles Weill subsequently joining xAI as a Member of Technical Staff.The company's failure reflects a structural mismatch between the capital required to sustain an ML-infrastructure-heavy product and the ceiling imposed by a deliberately narrow target market.

Image 1 / 2

Founding Story

Charles Weill founded CreatorML in 2022 under the legal entity Antifragile Research Inc., incorporated in New York. ^{[1] [2]} He came to the company as a solo founder — a deliberate choice, though the rationale for not recruiting a co-founder was never publicly stated.

Weill's credentials were exceptional for this specific problem. He spent seven years as a technical lead in applied machine learning at Google Research, working directly on Cloud AutoML, Ads, Search, and YouTube algorithms. ^[3] He also held an M.Eng in Computer Science from Cornell University. ^[4] Few people in the world had deeper institutional knowledge of how YouTube's recommendation engine actually worked — and Weill was building a product designed to help creators game exactly that system.

The founding insight did not come from a whiteboard exercise. It came from Twitter. Weill had been posting analyses of YouTube analytics data as a side interest, and he noticed that some of the platform's largest creators began following him. As he later described it: "I realized that I found a pain point when I began tweeting about simple analyses of YouTube analytics data, and many of the biggest YouTubers in the industry began following me." ^[5] This organic signal — major creators seeking out a random ML engineer's analytics posts — suggested genuine, unmet demand rather than a manufactured hypothesis.

Weill was also a part-time YouTube creator himself, which gave him first-hand exposure to the anxiety that drives creator behavior: the uncertainty of not knowing whether a title or thumbnail will perform before committing to it. ^[6] That dual perspective — insider knowledge of the algorithm and lived experience of the creator workflow — shaped the product's initial design.

The early period was financially precarious. Weill's LinkedIn activity references waking up to a $15,000 bill on his personal credit card while self-funding the startup, describing it as "the thanks I get for self funding my startup." ^[7] He bootstrapped through this period before securing institutional backing.