Designing for the Non-Human User
Aaron Levie, the chief executive of Box, stares at a future where your internet browser is largely obsolete. He believes software developers must stop building for human eyes and start building for the code that runs behind them. Digital architecture is currently undergoing a reorganization that favors autonomous agents over the people who technically own the accounts. Levie explained in a recent phone interview that the online world is gradually handing over its keys to these digital entities. He argued that agents could soon become the primary occupants of the internet, a transition that is already visible in the sheer volume of bot activity across enterprise platforms.
Software firms have historically prioritized user experience, focusing on aesthetic dashboards and intuitive menus. Such interfaces mean nothing to a large language model. Levie predicts that every piece of software will eventually require a total rebuild to accommodate these automated users. Silicon Valley is already responding. Labs like OpenAI, Anthropic, and Cognition AI are racing to deploy computers within computers that can scan legal documents, parse databases, and browse the web without human intervention. This transformation places a premium on backend efficiency rather than visual flair. If an agent can fetch the data, a human never needs to see the landing page.
Revenue models are also shifting. Levie anticipates a world where agents operate with independent budgets to solve business problems. An executive might assign an agent a twenty dollar budget to complete a complex logistical task. The agent would then interact with other software, pay for data access, and settle transactions on its own. While Bloomberg reports that some investors remain wary of giving credit cards to code, Levie remains convinced that bot-to-bot commerce is the next frontier of the global economy.
The logic is cold but commercially sound.
The Security Crisis in Beijing
Security concerns are complicating this rapid adoption in the East. China recently experienced a frenzy surrounding OpenClaw, an AI agent nicknamed the lobster by local netizens. Users rushed to install the software to automate mundane office tasks, but the trend has curdled into a security nightmare. Secondhand marketplaces like Xianyu, owned by Alibaba, are now flooded with listings for professional uninstall services. Technicians charge between 44 and 87 dollars to scrub OpenClaw from devices, promising to remove deep-seated files and potential viruses that the agent left behind. People who once celebrated the lobster are now paying a premium to kill it.
Regulators in Beijing have taken a dim view of this unauthorized automation. Government agencies and state-owned enterprises recently ordered employees to purge OpenClaw from work devices. Major banks have enforced similar bans, fearing that these agents might leak sensitive data or create backdoors for foreign intelligence. Employees must now report any previous use of the tool to their supervisors. The rush to delete the software has created a bizarre side hustle where experts visit homes to manually clean infected hard drives. Safety, it seems, has a secondary market price.
Cybersecurity analysts at various firms suggest the OpenClaw debacle highlights a massive vulnerability in the agent economy. When an agent acts on behalf of a user, it often requires administrative permissions that extend far beyond what a human would need. Once the agent is compromised, the entire system is open. Chinese social media platforms like RedNote are currently filled with warnings about residual code that continues to run even after the main application is deleted. These persistent processes can track keystrokes or exfiltrate files, turning a productivity tool into a permanent surveillance device.
Identity Theft by Algorithm
Creators in the West face a different kind of friction as platforms like Meta begin to automate human likeness. Julia Berolzheimer, a fashion influencer with more than one million followers, recently discovered that Instagram was using her images for a tool called Shop the Look. The feature used AI to identify products in her photos and offer them for sale without her consent or a commission. Berolzheimer described the experience as a form of theft, noting that her face and personal brand were being used to drive revenue for Meta while she was bypassed entirely. Many influencers now fear they are being turned into free training data for the very companies that host them.
Meta defended the practice as a limited test designed to help users find products they like. A spokesperson clarified that the company does not take a commission on these specific sales, yet the underlying message to creators is clear. Platforms own the digital real estate, and the people who fill it are becoming secondary to the algorithms that monetize it. TikTok and Pinterest are experimenting with similar tools. TikTok has even introduced AI avatars for marketers, suggesting a future where a brand can create its own influencer instead of hiring a human one. This shift threatens to dismantle the fragile economic pact that has sustained the creator economy for a decade.
Legislation has failed to keep pace with these developments. Copyright law was never designed to handle an environment where an AI can scrape a personality and repackage it as a sales tool in real time. Still, some creators are attempting to fight back by demanding new licensing agreements that specifically govern AI usage. If platforms continue to automate the value out of human content, the content creators may simply leave. But with agents now capable of generating their own media, the tech giants might not care if they do.
Technology is outpacing the legal framework meant to protect its subjects.
The Elite Tribune Perspective
Is there any greater irony than the fact that the tools built to liberate human productivity are now being optimized to exclude humans from the transaction cycle? We are watching the slow-motion eviction of the human user from the internet. Aaron Levie is right about the structural shift, but he is far too optimistic about what it means for the average person. When software is built for agents, it is built for speed and profit, not for transparency or human agency. The internet was once a bazaar of human ideas. It is rapidly becoming a high-frequency trading floor where bots negotiate with other bots over the scraps of our digital identities.
The OpenClaw crisis in China should serve as a wake-up call for anyone who thinks autonomous code can be easily managed. We are giving these agents the keys to our financial lives and our national security before we even know how to turn them off. Meanwhile, Meta is showing us that in the age of AI, your face is just another data point to be harvested for a click. This is not progress. It is the commodification of existence. If the future of the internet does not include a seat for the human at the table, then we are building a world that we no longer own.