In brief
- The creator of open-source AI agent platform OpenClawd said that he’s received acquisition offers from Meta and OpenAI.
- Peter Steinberger said he would only agree to a deal if the project can remain open source.
- He expects that AI agents can eliminate 80% of current apps, due to their versatility.
Peter Steinberger built the most important open-source AI agent in no time. Now Meta and OpenAI are circling with acquisition offers, he said last week, while crypto scammers turned his rebrand into a 24-hour nightmare that almost made him quit.
OpenClaw (formerly Clawdbot)—the self-modifying AI assistant that sparked MoltBook’s viral chaos and an entire ecosystem of autonomous agents doing increasingly weird and amazing stuff on the internet—hit 180,000 GitHub stars in record time.
The Austrian developer who “vibe coded” it into existence said he’s now choosing between billion-dollar corporate buyouts and staying true to the open-source ethos that made it explode.
“My conditions are that the project stays open source,” Steinberger told Lex Fridman in a three-hour interview for his podcast. “Maybe it’s gonna be a model like Chrome and Chromium. I think this is too important to just give to a company and make it theirs.”
Both Mark Zuckerberg and Sam Altman have made concrete offers, Steinberger said. Zuckerberg reached out via WhatsApp, and they spent 10 minutes arguing about whether Claude Opus or GPT Codex was better. Altman’s pitch came with something more tangible: A promise of computational power tied to the Cerebras deal that could dramatically speed up agent performance.
The project is currently hemorrhaging $10,000 to $20,000 monthly, he said. Steinberger routes all sponsorship money to dependencies rather than pocketing it. “Right now I lose money on this,” he said matter-of-factly, like someone who sold his previous company PSPDFKit and genuinely doesn’t care about the cash.
What he does care about is the name-change saga that nearly killed the whole thing. When Anthropic sent a trademark complaint over “Clawdbot” being too close to “Claude,” Steinberger renamed it to MoltBot.
Then crypto scammers struck. In the five seconds between pressing “rename” on two browser windows, bots sniped his accounts. They served malware from his GitHub. They hijacked his NPM packages. His Twitter mentions became unusable spam.
“I was close to crying,” Steinberger admitted. “Everything’s fucked.”
He almost deleted the project entirely. The second rebrand to OpenClaw required Manhattan Project-level secrecy, decoy names, and coordinating account changes across platforms simultaneously to avoid another crypto-scammer feeding frenzy. The attacks were so sophisticated that Steinberger called it “the worst form of online harassment I’ve experienced.”
Steinberger is also a fan of what Andrej Karpathy calls “agentic engineering”—a rejection of the term “vibe coding,” which he considers a slur.
“I do agentic engineering, and then maybe after 3:00 am, I switch to vibe coding, and then I have regrets the next day,” he explained. He runs four to 10 agents simultaneously, racked up 6,600 commits in January alone, and built most of the codebase by talking to AI rather than typing. “These hands are too precious for writing now,” he said.
Steinberger predicts that OpenClaw-style agents will kill 80% of apps.
“Every app is just a very slow API now, if they want it or not,” he told Fridman.
Why pay for MyFitnessPal when your agent already knows your location, sleep patterns, and stress levels? Why open Uber Eats when your assistant can order food, schedule meetings, and manage your calendar proactively?
Steinberger’s program has opened the gates to tech world giants. He told Fridman that he also held conversations with Microsoft CEO Satya Nadella. The dev is now weighing starting his own company with VC backing, but fears it would distract from building. He’s considered just continuing to bleed cash and ignore the offers.
“I can’t go wrong,” he said.
