Intelligence Raises $7.9M Seed Round as AI Feedback Platform Reaches $60M ARR

Artificial intelligence startup Intelligence has raised $7.9 million in seed funding to accelerate the growth of its human evaluation platform, Design Arena, which helps AI companies improve their generative models through large-scale user feedback.

The funding round was led by Index Ventures, with participation from Conviction, A*, Valkyrie, and several other investors.

From College Project to AI Evaluation Platform

According to co-founder Grace Li, the company began shortly before her graduation in 2025 when she and several classmates were building an AI-powered game engine.

While the models could generate playable games, the team struggled with a more difficult challenge—determining whether those games were actually enjoyable. They concluded that no automated benchmark could fully replace human judgment, prompting them to explore ways of collecting authentic user feedback at scale.

That idea evolved into Design Arena, a platform that enables users to compare AI-generated content and rank the results based on quality and preference.

"We realized it was the missing bottleneck for many of these models to improve in the design space," Li said, adding that the startup secured its first major contract with a frontier AI lab within a week of identifying the opportunity.

Design Arena Lets Users Rank AI-Generated Content

For consumers, Design Arena functions similarly to an advanced AI model comparison platform.

Users submit prompts through a ChatGPT-like interface and select the type of content they want to generate, including websites, images, and other visual formats. Instead of receiving a single response, they compare multiple AI-generated outputs in a series of "A vs. B" matchups before ranking the final results from best to worst.

These rankings create valuable preference data that AI developers can use to refine their media-generation models.

Human Feedback Drives Enterprise Demand

While the consumer experience helps attract users, Intelligence says the platform's primary value lies in its enterprise offering.

Because participants generally focus on selecting the highest-quality output rather than supporting a specific AI model, their rankings provide developers with relatively unbiased insights into user preferences.

According to Li, frontier AI labs are willing to pay for this continuous stream of human evaluation data, helping the startup reach $60 million in annual recurring revenue (ARR).

User Data Reveals Regional Design Preferences

The platform also collects additional insights by requiring users to log in before accessing generated content.

This allows Intelligence to analyze how design preferences vary across regions and evolve over time. Li noted, for example, that users in parts of Asia often favor more visually dense and maximalist web interface designs.

The company believes these behavioral insights complement automated AI benchmarks, which can be vulnerable to optimization strategies and manipulation.

Human Evaluation Market Continues to Evolve

The demand for human-generated evaluation data has grown as AI companies seek more reliable ways to measure model quality beyond automated testing.

However, success in the sector is far from guaranteed. Startup Yupp, which also focused on crowdsourced AI feedback, shut down earlier this year despite raising $33 million and attracting more than 1.3 million users.

At the same time, other companies have demonstrated strong momentum. LM Arena, which specializes in evaluating text-generation models through human comparisons, raised $150 million in a Series A round just months after launching its commercial product.

As frontier AI models become increasingly sophisticated, startups like Intelligence are betting that large-scale human evaluation will remain an essential component of developing AI systems that better reflect real-world user preferences.