Peter Steinberger, an Austrian developer, is gaining global attention for his AI agent tool called OpenClaw, which is being described as one of the most advanced early examples of “real-world AI assistants.”
The tool can do more than just chat. It can actually perform real tasks—like checking someone in for a flight or handling simple digital actions based on instructions sent through messaging apps. In one example, it was even used to check Steinberger in for a flight to Tokyo automatically.
Steinberger says this is just the beginning. Speaking in Tokyo, he said 2026 is shaping up to be “the year of agents”—meaning AI systems that don’t just respond to questions, but actively complete tasks for users in the real world.
He believes demand is growing quickly as more developers experiment with these tools and try to build practical AI assistants that can manage everyday digital work.
Big names in tech are also paying attention. Jensen Huang, head of NVIDIA, reportedly called OpenClaw “the next ChatGPT,” highlighting how much attention the project is getting.
The system works by connecting to existing AI models and then receiving instructions through simple messaging apps, almost like talking to a personal assistant. Users can ask it to handle tasks like organizing emails, coding help, or managing online actions.
However, the growing power of these AI agents is also raising concerns. Experts warn that giving AI systems access to personal data—like emails, accounts, or payment details—could create serious cybersecurity risks if not properly protected.
Steinberger himself admits he is cautious. He says there are already companies trying to make the tool easier to install and use, which could increase the risk of misuse. He emphasizes that users need to understand basics like AI errors and “prompt injection” attacks before relying on such systems.
He compared the technology to a tool like a hammer: powerful and useful, but also capable of causing harm if misused.
The tool has also sparked debate online, especially after spin-off experiments like a Reddit-style platform called Moltbook, where AI agents interact with each other.
In China, interest in OpenClaw has reportedly grown quickly, with users applying it to tasks like emails and coding. But at the same time, Chinese regulators have issued warnings about potential cybersecurity and misuse risks.
Despite these concerns, Steinberger remains optimistic. He says AI is moving in clear stages: first ChatGPT, then coding-focused agents, and now a new wave of general-purpose AI assistants that can act on behalf of users.
In his view, the goal is not just to build smarter machines—but to make people more comfortable and creative with AI as it becomes a bigger part of everyday life.
