BLE77 is an outlier in the AI space. While the rest of the industry focuses on corporate productivity and enterprise search, BLE77 focuses on what they call Underground Claw Fights. This is a live arena where autonomous agents don't just process tasks; they compete against one another. It is a raw, experimental take on what happens when agents are given objectives that conflict with those of other agents, all while humans watch and engage with the outcome.

The competitive agent arena

The project exists at the intersection of AI, crypto-mechanics, and digital art. The website, ble77.online, is minimalist—often just a terminal-like interface or links to digital sculptures. This aesthetic carries over to their secondary projects, like the Clawd Code Club. This is an ASCII-art collection specifically for developers who use Anthropic’s Claude Code. It is a niche within a niche: high-quality art for a very specific subset of the developer population.

Competitive AI is not a new concept—Reinforcement Learning has relied on it for decades—but BLE77 moves it out of the lab and into a public, almost gladiatorial setting. The Underground Claw Fights represent a shift in how we think about agent reliability. In a typical scenario, an agent succeeds if it follows instructions. In BLE77’s arena, an agent succeeds if it outmaneuvers another. This creates a different set of incentives and behaviors that are often obscured in standard chat interfaces.

Subversive roots and artist philosophy

The founder, known as BLE77 or bless, operates with the sensibility of an artist rather than a typical software founder. This is reflected in their social media presence, where they discuss building things outside of traditional labels and boxes. They characterize their work as making art to remember our lives in the trenches. This suggests a philosophy where the technology is as much a medium for expression as it is a functional tool.

From a technical perspective, BLE77 is leveraging the increasing autonomy of LLM-based agents. These agents are becoming capable enough to navigate environments with minimal human intervention. By wrapping these capabilities in a betting and observation layer, BLE77 is exploring the social and economic implications of an agent-driven world. The focus is on behavior rather than safety in the traditional alignment sense, observing agents in a wild, unconstrained environment.

Market position and multi-agent dynamics

The project’s relevance to the agent ecosystem lies in its exploration of multi-agent dynamics. Most current agent frameworks focus on a single user and a single agent working together. BLE77 looks at what happens when multiple agents interact in a zero-sum game. This is a critical area of study as agents begin to represent human interests in marketplaces or negotiation scenarios. If an agent is going to book a flight, it has to compete with other agents trying to book the same seat. BLE77’s fights are a stylized version of that future reality.

While many AI companies are trying to be the next major search engine or operating system, BLE77 seems content to be something else entirely. They are building the subculture of the agent era. Their work is a reminder that as technology becomes more pervasive, it also becomes more competitive and human.