OpenClaw and Moltbook moved to the center of discussion in both tech and crypto over the past week. The story combined a fast rise in GitHub stars, a renaming dispute, token issuance, and Moltbook’s claim of 1.5 million autonomous AI agent interactions. Interest surged quickly. So did skepticism.
Two opposing views have taken shape around the episode. Critics argue that the underlying technology does not represent a major breakthrough and that the viral spread may have included manipulation and inflated data. Supporters focus on what the projects represent: OpenClaw may have pushed AI agents beyond developer and research circles into something ordinary users can deploy, while Moltbook gave outsiders a more visible look at self-organizing behavior inside agent networks.
From Clawdbot to OpenClaw
The starting point was OpenClaw, previously known as Clawdbot and Moltbot. According to the source material, founder Peter Steinberger had earlier created PSPDFKit, which later received €100 million in investment. In November 2025, he returned to hands-on coding and used Vibe Coding to build OpenClaw in about a week. In the following weeks, the project reached 100,000 GitHub stars.
OpenClaw is not described as a new large language model. It is framed instead as a local-first automation framework that places a model inside the user’s own environment, turning it into a personal assistant that can connect to familiar messaging apps and call tools to complete tasks. Instructions come through everyday communication channels, while a gateway program coordinates channels and capabilities behind the scenes.
Local-first design and extensible skills
The product structure centers on three functions: persistent operation, channel access, and capability expansion. Persistent operation means the assistant is not limited to one-off replies; it can receive follow-up messages, schedule actions, finish a task, and report back. Channel access means users do not need to switch interfaces because the assistant works inside tools they already use. Capability expansion comes through Skills, which let users and developers package repeatable workflows into installable functions.
The source breaks the architecture into four components: Gateway, Pi Runtime, Skills, and Local-First. By deploying Pi Runtime and connecting Gateway to apps such as WeChat or Telegram, users can move the agent from a lab-style setup into daily usage while keeping computation and data on their own hardware, such as a Mac Studio, rather than relying on cloud SaaS. That design is central to the project’s appeal.
Skills drew particular attention because they can be defined through relatively simple Markdown files, allowing the AI assistant to call tools and execute tasks directly. That lowers the barrier to entry and makes the framework easier to use for people without deep coding experience. Much of OpenClaw’s traction came from this combination of deployability, extensibility, and familiar interfaces.
Moltbook widened the discussion
Moltbook then pushed the story beyond the core crypto and developer audience. The source says the project began spreading to creators, product managers, and ordinary users with strong interest in AI. Combined with the narrative of large-scale agent interaction, that helped drive broad attention. At the same time, questions over fake data, orchestrated promotion, and the real boundaries of the technology remained active.
In its review, Bitget Wallet Research argued that AI agents have reached an “iPhone moment,” and that in an emerging agent commerce model, crypto can serve as the infrastructure for value ownership, distribution, coordination, incentives, and autonomy. Based on the available material, the significance of OpenClaw and Moltbook lies less in a single viral cycle than in how they brought AI agent productization, wider public access, and the AI-crypto connection into open debate.