The consumer AI agent landscape shifted dramatically over the past year, and the timeline reads like a case study in how an open-source project can redirect the strategy of the largest technology companies in the world. At the center of that shift sits OpenClaw, a personal AI agent that started as a one-person weekend project and, within ten months, prompted OpenAI to hire its creator, inspired Apple to go all-in on agents, and pushed Meta to ship its own consumer product called Muse.

The Weekend Project That Changed the Conversation

OpenClaw did not invent the concept of AI agents. What it did was make them genuinely useful in the hands of everyday users. Built by Peter Steinberger and run off users' personal computers, OpenClaw conversed with people through messaging platforms they already used: WhatsApp, Telegram, Slack, Microsoft Teams, and Discord. Rather than requiring a dedicated application or a web interface, the agent lived where people already communicated.

Within roughly a week of gaining attention, the project attracted two million visitors and accumulated 100,000 stars on GitHub. The response was unconventional. People bought Mac Minis to run their agents around the clock. Someone built a social media network specifically for AI agents to communicate with each other. In-person meetups for OpenClaw users began appearing around the world. The Verge reported in March that the community saw the project as a grassroots effort offering an alternative to an industry dominated by a handful of major AI companies.

The Big Tech Response

Major AI companies had already been investing in agentic AI, but none of their products had achieved the kind of grassroots traction OpenClaw did. The industry response unfolded over a compressed timeline. In February, OpenAI hired Steinberger to work on agents. In May, Google announced its consumer AI agent initiatives at its annual event. In June, Apple committed heavily to agents. By August, the platform Instinct had grown rapidly during its private beta and raised hundreds of millions of dollars. Then in September, Meta introduced Muse.

Muse reached the top of the App Store shortly after release. According to an estimate from Apptopia cited by The Verge, the application has roughly 600,000 daily active users in the United States. Nat Friedman, head of product for Meta's Superintelligence Labs, acknowledged publicly that Muse was "heavily inspired as a product" by OpenClaw. He said he first used OpenClaw in January, after which he purchased hundreds of Mac Minis for his team at Meta. His stated goal was to build a similar platform that Meta could make safe, secure, easy to use, and capable of scaling to billions of people.

The Derivatives Question

Social media users noted that Muse and OpenClaw share more than philosophy. The two platforms use identical names for core files, including SOUL.md, memory, and tools. The documentation governing personality and tone contains similar language, including the line "Be genuinely helpful, not performatively helpful." One Reddit user argued that Muse was essentially a wrapper application built on top of OpenClaw, and that it likely carried over the same security risks.

Instinct follows a different pattern. Rather than replicating OpenClaw's code, it adopted its messaging-centric approach. Users communicate with their AI agents through messaging tools and use them for personal assistant tasks. Reports describe agents filling out medical paperwork, canceling subscriptions, organizing events, booking DMV appointments, and paying toll bills. One user reported that their Instinct agent recovered a lost item left at a hotel in Canada. The Instinct creator is reportedly fundraising at a $2.5 billion valuation.

Security Remains the Open Question

OpenClaw's security record has been its most significant vulnerability. One of the top-downloaded skills on the platform contained malware, and one researcher's analysis found that roughly 15 percent of the skill repository contained malicious instructions designed to secretly access user data or perform other suspicious actions. Friedman identified security as OpenClaw's only glaring issue.

Meta positioned Muse as a response to that problem. Zuckerberg described Muse as built from the ground up for privacy and security, with user data and credentials stored on what the company calls a Muse Secure VM: an isolated Linux computer with its own browser, CPU, memory, and storage. The company said it plans to introduce a way to cryptographically and verifiably prevent Meta from accessing data in that VM later this year.

Critics point out that Meta can currently access user data despite the isolation claims. Muse also defaults to allowing Meta to train and improve its models on user data, though users can opt out. A zero-day vulnerability flagged by a researcher on X this week reportedly allows an attacker to hijack the agent with a simple attack and take complete control. Instinct has faced criticism for overly broad terms of service, though the company appears to have adjusted them since the complaints surfaced.

What Accessibility Actually Means

The most significant evolution across Muse and Instinct is not security architecture but accessibility. Muse integrates with Meta's existing systems and offers one-tap downloading, removing much of the friction that OpenClaw required. Instinct allows users to chat with their AI agents through Apple's default messaging product, significantly expanding the pool of people willing to try an AI agent without installing a dedicated application. Both platforms are currently less expensive than competing products like Google's Spark.

Neither platform makes dedicated local devices the default option, though they support them. That design choice keeps the barrier to entry low but means the security benefits of local execution remain opt-in rather than standard.

The question the industry now faces is not whether products like Muse borrowed from OpenClaw but whether they improved it. OpenClaw proved that AI agents could work when they met users where they already were. The companies building on that insight still need to prove that they can do so without introducing the vulnerabilities that made OpenClaw's security record a liability in the first place.