Yupp.ai Shuts Down After $33M Raise: The Rise and Fall of Crowdsourced AI

The company, founded by Pankaj Gupta and Gilad Mishne, announced the closure on Tuesday, less than a year after its launch. Yupp.ai allowed users to test and compare results from 800 different AI models, including those from OpenAI, Google, and Anthropic, for free. The goal was to generate anonymized data on user needs in AI for model creators to pay for.

The startup attracted 1.3 million users and collected millions of preferences monthly; it even had a leaderboard. Additionally, Yupp.ai had AI labs as customers, but did not achieve a strong enough product-market fit to survive.

The founders attributed the closure to the fact that AI models improved by leaps and bounds. The current model, used by companies like Scale AI and Mercor, involves hiring experts, such as PhDs, for reinforcement learning. Furthermore, the industry is moving towards AI built for and used by other AIs, reducing the need for consumer feedback.

Pankaj Gupta, CEO of Yupp.ai, wrote in a post on X that the landscape of AI model capabilities has changed dramatically in the last year and will continue to change rapidly. The future is not just models, but agentic systems.

The startup raised a $33 million seed round in 2024, led by Chris Dixon of a16z crypto. In addition, Yupp.ai received investments from over 45 angels and small investors. Among the investors were prominent figures such as Jeff Dean, chief scientist at Google DeepMind; Biz Stone, co-founder of Twitter; Evan Sharp, co-founder of Pinterest; and Aravind Srinivas, CEO of Perplexity.

Despite the funding, Yupp.ai was unable to adapt to the rapid evolution of the AI market. Gupta mentioned that some employees will join a well-known AI company, while others are looking for new opportunities.

The closure of Yupp.ai highlights the dynamic nature of the AI sector and the challenges startups face in an ever-evolving market. The company failed to achieve a solid product-market fit, in part due to the rapid improvement of AI models and the shift in industry priorities.

Yupp.ai's experience serves as a reminder that even with substantial funding and a team of renowned investors, success in the world of AI is not guaranteed. Agility and adaptability are crucial for survival in this constantly changing environment.

The future vision of AI, according to Gupta, leans towards agentic systems, implying that interaction with individual models might be less relevant in the future. This could have affected the viability of Yupp.ai's business model, which was based on user feedback on different models.

The transition towards agentic systems suggests a change in the way we interact with AI, where autonomous agents could make decisions and perform tasks on our behalf. This change could require new strategies and approaches for the development and commercialization of AI.