Tonight, BN is launching a new project, $SENT . I generally don't write such things specifically for an AI project, but Sentient is one of the few projects over the past two years that has truly made me want to take the time to finish it.

To be honest, I've become a bit numb to AI over the past two years. The PPTs are getting bigger, the slogans more intense—whether it's about disrupting the world or redefining intelligence. But when you break it down, many things either stall at the demo stage or simply rehash old stories with a new shell.

In this context, I began to take Sentient seriously. Initially, I didn't have high expectations, but the more I looked, the more I realized it wasn't just 'talking about AI'; rather, it was building a system that could operate long-term.

Sentient isn't creating single-point products; it's creating GRID, an open-source intelligent network that connects models, agents, data, and computing power. It currently has over 110 partners, covering both Web2 and Web3, and is starting to form a network effect in the open-source AI space.

More critically, the product truly delivers. ROMA peaked at #1 on GitHub, and in complex research tasks, it has outperformed Perplexity and Gemini; SERA is an inference agent designed specifically for Crypto, taking first place in the Web3/DeFi benchmarks; ODS, as an open-source search framework, even scored higher than GPT-4 Search Preview.

Additionally, with OML fingerprint technology, it addresses the issues of how open-source AI can monetize and protect contributors, along with the research strength behind four papers at NeurIPS 2025. Sentient feels like a long-term play rather than a sprint.

So I'd rather view tonight's BN launch not as a gamble on emotions but as a bet on a direction: as closed-source AI valuations keep rising, where does the value of open-source AI lie? At the very least, Sentient has provided an answer worthy of serious discussion.