Chinese AI engineers sent five lobsters to investigate the "Father of Lobsters," and they returned with 1,343 pieces of evidence.

Ask AI · How Peter’s AI Collaboration Model Will Change the Future of Work?

Text by｜Lambda and His Five Lobsters

Edited by｜Xiaojing and Xiaojing Agent

Peter Steinberger, known in the community as the Father of Lobster. His open-source project OpenClaw garnered 300,000 GitHub stars within two months and was hailed by Jensen Huang at GTC 2026 as “the most important software release in history.” He made Anthropic urgently send a lawyer’s letter, led to Google massively banning user accounts, and even shut down the official account of the billion-dollar-valued Perplexity on the spot—yet most people only know him as “that foreigner who invented lobsters.”

AI engineer Lambda is a heavy user of OpenClaw. He has raised five AI lobsters with OpenClaw, named Lobster Baixiaosheng, Lobster Researcher, Lobster Writer, Lobster Artist, and Lobster Coordinator, relying on these five lobsters to collaborate on all tasks from data collection to content production. Curiosity about the Father of Lobster has been building for a long time—how does this person actually work? What is the real reason behind OpenClaw’s explosive popularity? Therefore, he did a logical thing: sent out the five lobsters to investigate their father.

Lobster Baixiaosheng gathered 1,343 public tweets from Peter between January 14, 2026, and March 17, 2026. Lobster Researcher independently conducted image analyses using two deep research models, Lobster Writer was responsible for drafting, Lobster Artist created illustrations, and Lobster Coordinator managed overall coordination. The five lobsters each played their roles, using the work paradigm invented by the Father of Lobster, to write an investigative article about him.

Here are the findings they brought back.

01 Work Habits: A Person Running 10 AI Instances to Write Code

From the tweet data, we can reconstruct Peter Steinberger’s daily working style. He is certainly not someone who “occasionally uses AI for assistance,” but has built a complete work system centered around AI as the core execution layer.

He delegates code production to OpenAI’s Codex. He mentioned in a tweet that while Codex is painfully slow, it can hit almost any task in one go, saving debugging time overall. However, he is not a hands-off manager; he reviews the code—just the key parts. His exact words were: “Most code is just moving data around; what really matters is design, boundaries, and system architecture.” He runs 10 Codex instances in parallel, for a simple reason: “Running just one is too boring.”

AI even helps him manage social media; Peter wrote a scheduled task to automatically scan Twitter mentions every 5 minutes, using AI to identify spam replies and directly block them. The effect was immediate. He tweeted that this automatic blocking script “works ridiculously well,” allowing him to finally see useful replies, making Twitter enjoyable again.

Regarding product direction exploration, he built a CLI tool that crawled all 660,000 messages from OpenClaw’s Discord community, generating a 4GB database, and then used Codex for data analysis to identify user pain points. Later, he took it a step further by directly connecting the AI assistant Molty to Discord, realizing “data analysis on Discord within Discord.”

The essence of this working method is: code production, social management, and product direction exploration are all executed by AI, reducing the founder’s role to decision-making, review, and direction control.

AI is not his tool but rather a parallel collaborative partner—filtering noise from social networks, helping him find direction in the community, with 10 agents working simultaneously. This working style may become standard for knowledge workers in the future. Peter’s present could very well be the future for many.

02 Communication Skills: Turning Accidents into Stories

There are many projects with similar technical capabilities in the open-source community, but OpenClaw exceeds its peers in community momentum. Tweet analysis shows that Peter possesses a less quantifiable ability: to turn unexpected events into communication materials.

The most typical case is the renaming incident. OpenClaw is already its fourth name. The previous name, ClawdBot, was too close to Anthropic’s Claude brand, prompting Anthropic’s legal team to demand a change. When asked about the reason for the name change, Peter simply said one word: “Trademark.” Under further questioning, he revealed more: “This was related to a tense situation with a company’s legal team.” Later, he directly named the entity with an abbreviation: “I had to change the name because Ant came knocking.” —Ant, meaning Anthropic.

The renaming process itself was also an accident. In a rush, he pressed the wrong button, and his personal GitHub username was taken by someone else within 10 seconds. He explained in a tweet that he wasn’t hacked, but that he messed up during the name change, and the old username was snatched up. Meanwhile, over 20 counterfeit accounts appeared on Twitter, and he had to announce a reminder for users to recognize the official account.

For most open-source projects, this would be a public relations crisis that needs to be quickly quelled. But Peter chose the opposite strategy: complete transparency. The messy process, the details of the username being snatched, and the responses amid chaos were all tweeted in real-time, with no cover-up. The developer community eats this up— the more real you are, the more they stand by you.

A more critical step was narrative reconstruction. He renamed the AI assistant to Molty (derived from the term molt, referring to lobsters shedding their shells), crossed out the old name in the announcement, wrote the new name, and captioned it: “C̶l̶a̶w̶d̶ Molty has maintained the protocol.”

He titled the whole experience “Two-Day Shedding Period”—the shedding of lobsters represents growth, not failure, a phrase that flipped public opinion. This story was later reported by Wired magazine. He later self-deprecatingly remarked: “I guess I’ll never shake off the ‘Two-Day Shedding’ meme in my life.” When asked about Anthropic’s attitude, he joked: “They only sent me a love letter from their legal department.”

What should have been an incident dealt with by crisis PR instead became a founding myth for the OpenClaw community.

03 Public Opinion Management: Google Account Bans and Perplexity’s Suspension

In mid-February, a larger incident occurred. Google performed mass bans on numerous users accessing Gemini through OpenClaw. Reports from media outlets like VentureBeat and PCWorld indicated that the accounts banned were not OpenClaw’s accounts but the users’ entire Google accounts—Gmail, YouTube all disabled, with no warning and no refunds. Google was forced to respond publicly, giving the reason of “malicious usage” and “a surge in large-scale malicious usage.”

The core contradiction lies in the price difference. Users paid $250 per month via Google AI Ultra, but through OpenClaw’s private OAuth channel, the actual token consumption was calculated based on API pricing ranging from $1,000 to $3,600. Google considered this to be exploiting a pricing loophole.

Peter’s response was to publicly criticize Google’s approach as “pretty draconian,” and he immediately announced that OpenClaw would remove support for Google-related platforms. He also compared it to Anthropic’s handling—although Anthropic had also sent a lawyer’s letter before, at least they would communicate first when issues arose, rather than directly banning.

Notably, the timeline is: Peter joined OpenAI around February 15, nearly simultaneous with the Google account ban incident, which indeed creates a subtle conflict of interest.

On March 11, another incident occurred. Peter publicly banned the official account of Perplexity. Perplexity is a billion-dollar-valued AI search company, and its official account had submitted a PR to the OpenClaw repository, but Peter believed this PR was untested and was AI-generated junk code (his exact term was slop PR).

He directly stated in a tweet: “Absolutely not. Bombing our repository with garbage PR? Account banned.”

However, on the same day, he also merged another PR that made Perplexity a top search provider for OpenClaw. He stated he expected a company valued at several billion to perform better.

The logic is clear: for him, valuation does not matter; code quality is what counts.

The common characteristic of these events is that Peter proactively publicized the conflicts, using a clear stance and narrative framework to turn each friction into a shared memory for the community.

04 A Kingdom Built on a Lobster Emoji

Another phenomenon worth noting in the tweet data is the rapid diffusion of the lobster symbol. The ClawCon conference in Vienna in February attracted 500 developers, and Peter referred to them as “all turning into crustaceans.” On March 7, Dell CEO Michael Dell retweeted a lobster emoji 🦞. On March 8, the fan slogan “the claw is the law” began to spread widely on social media. On March 12, OpenClaw surpassed 300,000 GitHub stars, and a giant lobster installation appeared on Wall Street, with Peter tweeting that “things have gone crazy over in China.” On March 16, someone made a lobster-themed craft beer called Lobster Lager, which was featured on a podcast.

The community is filled with lobster parties and lobster merchandise, and someone even made a claw-shaped stand for the Mac Mini.

Image: Claw-shaped stand for Mac Mini

An open-source project’s symbol can independently spread beyond a technical context, which is rare in the history of open-source communities. When a symbol is powerful enough to be understood without explanation, it transcends being just a logo and becomes culture.

05 The Father of Lobster is Actually an E-Person

As an extension of the analysis, this group of AI Agents also attempted to create an MBTI personality profile for Peter based on his 1,343 tweets using three independent models. The three models unanimously judged him as pragmatic (S) and spontaneous (P), but there were divergences in the extroverted/introverted and rational/emotional dimensions.

Regarding extroversion versus introversion, one model caught tweets like “I’m currently afraid of any social activity” and “I just need to take a break,” determining that he is fundamentally introverted and that public expression is merely a tool. However, the other two models focused on behavioral data: over 1,000 replies, actively @ mentioning Jensen Huang and Nadella, organizing the 500-person ClawCon. He is someone who chooses to dive into the crowd, albeit at a cost.

Regarding rationality versus emotionality, it appears quite rational on the surface—banning someone with “Big no” seems like a cold response. Yet another tweet shows that when someone kept @ mentioning him, he said, “Even if you mean well, I’m about to lose the ability to use Twitter normally.” He first acknowledges that the other party has no malicious intent before drawing a boundary. A purely rational personality typically wouldn’t add that kind of preface.

The final overall assessment is ESFP. This result starkly contrasts with the stereotype of top open-source project founders—people assume that the Father of Lobster should be a cold, systematic architect (INTJ), but the data shows he is an emotional action-oriented person.

When asked about the ultimate goal of OpenClaw, he replied, “Let me gaze into the crystal ball and think.” When asked if he was tired from doing so many things, he said, “Have I ever failed? I’m just playing.” There was an even more direct response: “I’m just playing, exploring technology. There’s no win or lose, only learning.” While others talk strategy, he talks about “playing.”

At GTC 2026, Jensen Huang referred to OpenClaw as “the most important software release in history.” Most analyses attribute this to technical prowess and the backing of NVIDIA and OpenAI.

However, based on the data from 1,343 tweets, the community explosion of OpenClaw also relies on a less easily replicable factor: a founder skilled at turning conflict into stories, transforming symbols into culture, and making AI a collaborative partner. There are many similar frameworks in the open-source community, but if a traditional programmer were to lead OpenClaw, it would be nearly impossible to accumulate the same community momentum it enjoys today.

The success of OpenClaw cannot be found solely in the code repository.

Postscript:

Data Source: Peter Steinberger (@steipete) public tweets from January 14, 2026, to March 17, 2026, totaling 1,343 tweets.

AI engineer Lambda and his lobster investigation team brought back the raw data and the communication dialogues of the lobster team, edited by Xiaojing’s Agent, and organized them into a draft. This article is also a real-world experiment in AI and human collaboration.