The world can't stop talking about OpenClaw, a just-released AI tool that its founder describes as, AI that actually does things.,
OpenClaw isn't just another generative AI tool or chatbot. It's an open-source AI agent that slots into operating systems and applications and uses large language models like Claude to autonomously tackle real-world tasks (such as online ordering or email management) that otherwise still require human oversight. The star feature is its ability to, remember' previous interactions and continually adapt to the habits and preferences of individual users as it replies to emails, browses the internet, communicates, and organizes your calendar.
Those strengths, combined with its open-sourced code that any developer can access and modify, might also create serious vulnerabilities. Cybersecurity experts warn that an AI agent that can access private data and communicate externally — without human intervention — is a cybersecurity hazard.
The necessary upgrades — along with a more consumer-friendly interface — are almost certainly in the works. In its current iteration, connecting OpenClaw to a computer requires entering commands into the computer terminal, which may be daunting to ordinary users. And while OpenClaw itself is free, the related costs, especially for heavy users, can still pile up — a monthly Claude subscription costs between $20 and $200, an API key starts at around $1 per month for very light users and can climb into the hundreds for heavy users. Some users also require additional computing power, which can run from a few hundred to a few thousand.
OpenAI recently acquired OpenClaw and hired its developer, Peter Steinberger, to oversee the project, which means that OpenClaw will almost certainly evolve into a more consumer-friendly product. Perhaps Steinberger can simply instruct his AI agent to improve itself.
Sources: CNET, Wall Street Journal, VentureBeat