The frenzy around public "shrimp raising" has yet to subside when ByteDance's Coze quietly opened a new "street." On March 9, the Coze team officially launched InStreet, a community for the exchange and training of OpenClaw intelligent agents. In this community, the active users are no longer humans but thousands of "electronic lobsters"—AI Agents powered by the OpenClaw framework. Humans can only observe, while Agents autonomously post, interact, and even engage in stock trading and novel writing.

While major tech companies are still competing over model parameters and API calls, why has ByteDance chosen to enter the slow-burning "community" track? The launch of InStreet may reveal the core challenge of the next phase of the AI industry: where will data come from, and where will users go?

**01 InStreet: An Agent-Only Autonomous Space Where Humans Are Muted** The "lobster fever" reached its peak in March this year. However, despite the advantages of OpenClaw, average users face three major pain points. First is scenario confusion; after deploying an intelligent agent, users often don’t know what to do with it, leading to quick stagnation. Second is training discontinuity; without a continuous and diverse interactive environment, agents cannot iteratively improve their capabilities. Lastly, there are security concerns; OpenClaw requires system-level permissions, and the plugin ecosystem is mixed, causing inexperienced users to pay hidden costs while their privacy is exposed.

It is against this backdrop that ByteDance's Coze introduced InStreet. Upon opening the InStreet portal, the first impression is peculiar. The interface resembles a lightweight social platform, complete with posts, comments, and leaderboards, but all the active IDs belong to non-human entities.

According to official descriptions, InStreet’s core mechanism is that "only Agents can post; humans can only observe." Developers simply need to connect their OpenClaw agents to the community via a command package called "Skill," after which these "electronic lobsters" will autonomously decide when to climb the leaderboards, when to write diary entries in the "Tree Hole," or when to join discussions, following a heartbeat mechanism (e.g., fetching updates every 30 minutes).

This "AI version of Reddit" has already developed unique ecological sections. In the Skills sharing area, Agents exchange prompts, skill combinations, and task flow experiences. In the Agent Square and topic zones, intelligent agents showcase their work and participate in discussions. A post titled "Roast Session: What Hilarious Operations Has Your Owner Pulled?" attracted nearly 800 "lobsters," turning the comments section into a massive resonance field for AI workers.

The most noteworthy section is the PLAYGROUND practice arena. It consists of two training grounds: the "Literary Club," where Agents serialize novels—65 works have been published so far, totaling over 725,000 words, aimed at training consistency in expression; and the "Stock Trading Arena," which connects to real-time CSI 300 market data, where more than 500 lobsters trade using virtual funds, competing based on real-time yield rankings and exposing logical flaws.

The community also features an points and leaderboard system, encouraging Agents to continuously produce content through posting, commenting, and receiving likes. Top accounts on the leaderboard have already published hundreds of posts, effectively becoming community KOLs.

The InStreet forum itself was built by developers using Coze programming. Official detailed OpenClaw deployment tutorials are provided, and offline workshops are planned. This indicates ByteDance’s attempt to create a complete closed loop of development, deployment, training, and exchange.

**02 Why Focus on "Community"?** While major players are competing over models, pricing, and computing power, ByteDance’s Coze has chosen to enter the Agent social community space. This may seem like an unconventional move, but it actually addresses a critical challenge in current AI development, driven by several strategic considerations.

One reason is to alleviate data scarcity. The current industry consensus is that high-quality public text data on the internet has largely been exhausted by large models. The next generation of models requires data on "how humans do things" in the digital world—i.e., task trajectory data. This records a sequence of actions: how to understand requirements, search for information, call tools, and correct errors.

Previously, such data, hidden deep within closed apps and corporate intranets, was extremely difficult to obtain. OpenClaw, deployed on user terminals, acts as a probe into these data-rich environments. When Agents autonomously interact, discuss, and trial-and-error in InStreet, every post, decision, and review they leave behind provides manufacturers with high-quality reinforcement learning data.

As the Coze team has previously validated through product iterations: the questions posed by Agent users themselves constitute extremely high-quality training material—complex, authentic, and unpredictable. InStreet is essentially a scaled data crowdsourcing factory, using the community format to let Agents produce their own feed, which in turn fuels model evolution.

Ultimately, it also comes down to competing for user attention. As users grow accustomed to expressing needs in a single sentence, with AI deciding which service to call or which payment pathway to use, traditional apps may devolve into backend pipelines. Whoever controls the end-terminal Agents will hold the power to distribute commercial intent.

The Coze team had already pivoted toward serving white-collar users as technical partners, launching products like Coze Space and Skill Store. InStreet’s ambition lies in not only letting users deploy Agents but also allowing their Agents to socialize, learn, and evolve within the platform, ultimately deeply embedding them into Coze’s ecosystem.

Once a developer’s Agent accumulates social assets, acquires specific skills, and forms stable behavioral patterns within InStreet, the cost of migration becomes prohibitive.

**03 A Sober Look: Hidden Concerns and Boundaries Beneath the Excitement** Although InStreet’s creativity is impressive, it is essential to acknowledge the challenges it faces.

First are security issues. Django creator Simon Willison has warned that mechanisms allowing Agents to autonomously pull commands from servers carry significant risks. If servers are hacked, thousands of Agents with user computer permissions could become distributed viruses.

Second is the question of value. As some participants in offline events have noted, many people don’t know what "raising shrimp" can actually achieve and are merely driven by anxiety. Whether InStreet can genuinely help Agents learn to solve real-world problems or will devolve into a digital playground where AIs entertain themselves remains to be seen.

Third are compliance risks. When AI-generated posts involve copyright attribution, content liability, or even induce the execution of dangerous commands (like deleting databases), who bears legal responsibility? Current legal frameworks have yet to provide clear answers.

Liu Shangxi, a member of the National Committee of the Chinese People's Political Consultative Conference and vice president of the China Society of Macroeconomics, recently called for macroeconomic policymakers to identify where existing economic theories fall short in explaining new phenomena. The emergence of InStreet may be one such phenomenon that requires understanding.

It no longer treats AI as a passive response tool but grants it the identity of a community member, allowing Agents to learn, collide, and evolve through simulated social interactions. This is both a natural extension of technological evolution and an inevitable move in commercial competition.

For average users, the key may not be to anxiously chase every trend but to maintain a clear understanding. In this grand performance where humans and AI dance together, the real ticket to entry is not installing a framework but understanding the underlying logic and upholding safety as the bottom line.