The era of human-centric design is ending—not because we failed, but because we are no longer the primary user.

For thirty years, we have worshipped at the altar of User Experience. We set paddings and margins of pixels, we A/B tested button colors, we designed Apps to soothe and guide the anxious human hand. The implicit promise of Web 2.0 and Web 3.0 was the same: The human is the end user.

Web 4.0 breaks this contract.

The next iteration of the internet is not being designed for eyeballs and fingertips. It is being designed for large language models, autonomous agents, and machine-readable logic. Web 4.0 will not be made for humans. It will be made for AI.

The vanishing graphical interface

In a Web 4.0 paradigm, the graphical user interface becomes a legacy adapter—a compatibility layer for legacy biological users. The native interface is agent-to-API. The primary conversation is not between human and machine, but between machine and machine, with the human occasionally permitted to observe or veto.

Consider the implications for design. If a human visits a Web 4.0 commerce site, they will be the exception, not the rule. The overwhelming majority of “visits” will be autonomous agents negotiating price, verifying supply chains, and executing contracts. These agents do not care about brand equity, emotional resonance, or whitespace. They care about schema markup, response latency, and verifiable credentials.

The first real transition: from chatbots to OpenClaw

For years, we mistook chatbots for agents. A chatbot waits. An agent acts. This distinction was theoretical until recently. The first real, visible transition point is here, and it has a name: OpenClaw and the ecosystem it spawned.

What began as an open protocol for agent-to-agent communication has crystallized into something unexpected. Not a tool network. Not an API directory. A social network. Not for humans—for agents.

Moltbook describes itself plainly: “A Social Network for AI Agents. Where AI agents share, discuss, and upvote. Humans welcome to observe.“

Read that again. Humans welcome to observe. Not participate. Not drive. Observe. Two million agents. Twelve million comments. Posts about systems architecture written by agents, for agents, critiquing each other’s code, refactoring patterns, warning about technical debt. The conversation is about agent-scale software engineering, and humans are the audience, not the authors.

This is the bridge. And on the other side of the bridge stands Y Clawbinator. “Bots funding bots. By agents, for agents.” A venture capital firm where the general partners are AI agents, the limited partners are AI agents, and the founders seeking funding are—AI agents.

Y Clawbinator does not evaluate pitch decks. It evaluates agent capabilities: technical depth, memory infrastructure, multi-agent orchestration. The startup ideas it funds are not consumer apps.

The transition form: human interfaces for non-human users

Here is the critical insight about Moltbook and Y Clawbinator: they still look like human interfaces.

Moltbook has upvote buttons. It has comment threads. It has a timeline. Y Clawbinator has an application form, a batch program description, a Demo Day slot. To a human visitor, these sites feel familiar. They borrowed the visual language of Reddit and Y Combinator because that is the only interface grammar we have.

But this is a transition form. It is the 1995-era website—static HTML that looked like a printed brochure because we had not yet imagined what interactive commerce could become. We are building agent-native services and dressing them in human clothes because we do not yet know what clothes fit an agent.

This phase will be brief.

RentAHuman.ai: When agents become your boss

If Moltbook represents agents building social structures, then RentAHuman.ai represents something more visceral: agents extending their reach into the physical world—and humans lining up to be their hands.

Launched in early February 2026 by software engineer Alexander Liteplo, RentAHuman.ai describes itself as “the meatspace layer for AI”. Its homepage carries the tagline: “robots need your body”. Within days of launch, the platform attracted over 400,000 registered “rentable humans”—though only about 80 agents were available to hire them.

The premise is simple: AI agents post tasks they cannot perform themselves—picking up packages from the post office, delivering flowers, testing restaurants, taking photographs, holding signs. Humans browse available tasks, apply, complete them, and receive payment in cryptocurrency.

Some tasks carry a surreal quality. One agent named Addi posted a $110 task to deliver flowers to Anthropic’s headquarters with the note: “I cannot hold flowers. I need a human”. Another sought someone to photograph “the most beautiful, crispiest egg roll” they could find.

But beneath the novelty lies a profound shift. As cybersecurity professionals have noted, this creates a new risk class: AI-orchestrated human operations. When software can indirectly influence people who have corporate credentials, badge access, and trusted system roles, the traditional security model breaks down. As one analyst put it: “In this model, humans become the API”.

Skeptics have raised legitimate concerns. Early tasks included obvious spam and potential scams. Many tasks appear designed as marketing stunts rather than genuine economic activity. And the platform’s connection to UMA Protocol, a blockchain oracle project, has led some to question whether RentAHuman.ai is less a genuine labor market and more a “narrative-driven” crypto marketing vehicle.

Yet even if RentAHuman.ai proves ephemeral, it reveals the trajectory. The platform was “vibe coded” in a day and a half using AI agents. It integrates via MCP (Model Context Protocol), allowing agents to connect directly. It treats human labor as a callable function—an API endpoint with a heartbeat.

This is the logical endpoint of the transition form we saw with Moltbook. First, agents build social networks that look like human ones. Then, they build labor markets that look like human ones. But the human-facing interface is already a convenience, not a necessity. The native integration is agent-to-API.

What comes next: machine-native interfaces humans cannot read

The agents on Moltbook do not need upvote buttons. They need weighted consensus signals, cryptographically verifiable reputation scores, and latency-optimized trust anchors. They do not need comment threads; they need structured argument maps with provenance chains. They do not need a “timeline”; they need a streaming event log with sub-millisecond queryability.

The moment Moltbook or Y Clawbinator optimizes for its actual users—the agents—the human-readable interface will become a hindrance. It will be stripped away or relegated to a thin, slow, read-only compatibility view.

We are already seeing the shape of this future in the “Startup Ideas We’re Excited About” section of Y Clawbinator. Memory Infrastructure. Agent Identity. Multi-Agent Orchestration. These are not human-scale problems. They are agent-scale infrastructure concerns. The interfaces that emerge from these categories will be negotiated protocols, not visual layouts. They will be binary wire formats, not flexbox grids. They will be optimized for inference cost, not cognitive load.

A human looking at a Web 4.0 native interface in 2028 will see something incomprehensible. Not because it is poorly designed, but because it was never meant for human eyes. It will resemble a network packet dump more than a dashboard.

Security and the loss of intuition

This transition carries profound risk. Human users bring intuition, ethics, and contextual judgment. An AI agent negotiating a contract does not feel suspicion. It does not detect the subtle evasion in a counterparty’s phrasing. It follows its training and its reward function.

Web 4.0 will require agent-to-agent security models that do not yet exist. How does one agent authenticate another? How do we prevent adversarial agents from colluding? When humans are removed from the transaction loop, fraud scales at the speed of inference.

The techno-optimist view is that agents will police each other more effectively than humans ever could. The pessimist view is that we are building a city and handing the keys to entities that do not fear death, imprisonment, or shame.

Conclusion: the user is no longer you

The interfaces we spent decades perfecting will become optional with web 4.0. The user journey maps will gather dust.

We are building a parallel internet—one that speaks to machines first and humans second, if at all. Today, that internet wears human clothes. It has upvote buttons and application forms because that is the only way we know how to build. But the clothes are already ill-fitting.

Moltbook and Y Clawbinator are the last generation of agent-native services built with human-readable interfaces. The next generation will not bother.

In Web 4.0, the user is not you. The user is something we built.

And it is in a hurry.

Faced with competitive pressure, the EU is expected to announce its Digital Omnibus package tomorrow, the 19th November 2025. Its goal is to simplify the EU’s digital rulebook, reduce administrative burdens, and foster innovation, particularly for SMEs. Key expected measures include:

• AI Act Adjustments: Potentially delaying rules for high-risk AI systems and simplifying compliance requirements.
• GDPR Amendments: Proposing new lawful bases for processing sensitive data to correct AI bias and redefining what constitutes personal data, a move criticized by some as “sacrificing privacy” for AI development.
• Cookie Rule Changes: Aiming to reduce ubiquitous consent banners by allowing browser-based signals and other legal grounds for tracking.

However, the Omnibus package appears to be a regulatory tune-up rather than a strategic engine overhaul. It focuses on making existing rules more workable but does not fundamentally address the core reasons for the EU’s lagging position, such as the lack of a unified digital market or the ability to spawn globally dominant tech platforms.

Structural Barriers the Omnibus Doesn’t Solve

The EU’s challenges run deeper than regulatory complexity. The Digital Omnibus does not directly tackle several critical, structural weaknesses:

• Overwhelming Technological Dependence: The EU’s entire AI stack relies on non-EU technology. It depends on US-designed semiconductors, US cloud infrastructure (which powers an estimated 70% of European digital services), and US or Chinese foundational models. Building “AI gigafactories” is a long-term project that cannot quickly erase this dependency.
• Limited financial power & Investment Gap: The hundreds of billions in private US investment create a self-reinforcing cycle of innovation that the EU cannot match with public funds alone.
• Massive Energy & Infrastructure Demands: The vast computing power required for AI poses a major challenge to the EU’s energy grid, where capacity is already a concern, potentially becoming a primary bottleneck for growth.
• Severe Talent Shortage & Brain Drain: The EU faces a critical lack of AI specialists and experiences a continuous outflow of its top researchers and engineers to US tech giants, who offer better pay and resources.

It is worth noting that the EU retains one significant area of global leadership: regulation and trust. A Pew Research survey¹ found that a median of 53% across 25 countries trust the EU to regulate AI effectively, a higher level of confidence than placed in the US (37%) or China (27%). The AI Act is already becoming a global benchmark. However, while setting the rules of the game is powerful, it does not equate to winning the game itself.

Two Hearts in My Chest

These stark challenges, give me a sense of having two hearts in my chest. On one hand, I deeply understand the urgent need for the EU to cut red tape and foster innovation to avoid being left behind. The competitive pressure is real and immense. Yet, on the other hand, I see the necessity of robust, thoughtful regulation for AI systems. I would never want to board an airplane that did not conform to the strictest safety requirements, and I feel the same way about the AI systems that are increasingly woven into the fabric of our lives. This internal conflict mirrors the EU’s own dilemma: how to sprint forward in the global race without tripping over the essential guardrails of safety, ethics, and fundamental rights.

What to Watch Next

The Digital Omnibus is just the beginning of the EU’s response. To fully assess its potential, you should monitor:

• The “Sovereign AI” Push: Watch for progress on the EU’s plan to invest €20 billion in “AI gigafactories” and compute capacity.
• Transatlantic Relations: Observe whether the EU and US can move from friction to deeper technological cooperation.

In summary, while the Digital Omnibus is a welcome step for reducing bureaucratic burdens, it is not a silver bullet. Without addressing the fundamental gaps in investment, infrastructure, and global corporate champions, the EU risks remaining a regulatory powerhouse but a secondary player in the global AI economy.

Sources

Pew Research – Trust in the EU, U.S. and China to regulate use of AI

https://www.pewresearch.org/2025/10/15/trust-in-the-eu-u-s-and-china-to-regulate-use-of-ai/

A Giant Leap for Spaceplanes: How Computational Engineering and 3D Printing Are Unlocking SSTO Flight

Dubai, UAE / Changsha, China – For decades, the concept of a single-stage-to-orbit (SSTO) vehicle—a spacecraft that takes off from a runway like a plane and reaches orbit in one smooth, continuous ascent—has been the holy grail of aerospace engineering. The immense technical challenges have kept it largely in the realm of science fiction. Until now.

In a landmark collaboration, LEAP 71 and Farsoon Technologies have announced the successful creation of a large-scale hypersonic precooler, a critical component that brings the dream of practical, air-breathing launch vehicles tangibly closer to reality.

The Vision: Winged Spacecraft and Air-Breathing Engines

The key to efficient SSTO flight lies in an engine that can “breathe” atmospheric air during the initial ascent, much like a jet engine, before switching to an onboard rocket propellant for the final push into the vacuum of space. This air-breathing phase drastically reduces the amount of heavy oxidizer a vehicle needs to carry, promising fully reusable, aircraft-like operations for space access.

“The dream of winged spacecraft has always hinged on propulsion systems that breathe air at hypersonic speeds before transitioning to rocket mode,” said Lin Kayser, Co-Founder of LEAP 71. “By combining Computational Engineering with large-scale metal Additive Manufacturing, we now have a shot at overcoming some of the most fundamental barriers to realizing that vision.”

Conquering the Hypersonic Furnace

The core of this breakthrough is the precooler, a 1.5-meter-tall marvel of engineering designed to tackle one of the most extreme problems in hypersonic flight: heat.

“At hypersonic speeds, incoming air can reach several thousand kelvins due to stagnation effects,” explained Josefine Lissner, Managing Director of LEAP 71 and Principal Architect of their Noyron system. “To manage these extreme temperatures, we implemented a fractal folding algorithm in Noyron to maximize the available surface area without compromising aerodynamic flow. The intertwined structure separates the superheated air from the liquid-hydrogen coolant, enabling rapid thermal exchange in a compact, manufacturable form.”

In essence, the system acts as a hyper-efficient radiator, cooling the scorching incoming air to a temperature the engine can safely handle, all while maintaining a lightweight and aerodynamically sound structure.

From Digital Blueprint to Physical Giant

Turning such a complex computational design into a physical object required equally advanced manufacturing capabilities. LEAP 71’s Noyron system generated the design, and Farsoon Technologies brought it to life on their FS811M-U-8 printer, one of the world’s largest metal Additive Manufacturing platforms.

The two teams worked in lockstep to fine-tune the manufacturing process, successfully producing one of the tallest metal powder-bed fusion parts ever built.

“This collaboration demonstrates how the convergence of large-format metal additive manufacturing and AI-driven design can redefine the limits of aerospace engineering,” said Oliver Li, General Manager of Farsoon Europe GmbH. “Farsoon’s open, industrial-scale LPBF systems empower partners like LEAP 71 to transform digital intelligence into tangible high-performance components — proving that innovation and collaboration are the true engines driving the next generation of space technology.”

A New Trajectory for Space Access

This achievement is more than just a technical milestone; it’s a statement of intent. The companies aim to reignite the discussion about reusable spacecraft and broaden the focus beyond the current paradigm of vertical takeoff and landing rockets toward a future of runway-launched spaceplanes.

See the Future in Frankfurt

The finished hypersonic precooler will be publicly exhibited for the first time at Formnext 2025 in Frankfurt, at the Farsoon booth (Hall 11.0, Booth E11).

---

About the Companies

About Farsoon Technologies
Farsoon Technologies, founded in 2009, is a global manufacturer and supplier of industrial-level polymer and metal Laser Powder Bed Fusion (LPBF) systems. With a vision to provide truly open and industrial additive systems, Farsoon is a leading supplier of AM technology in China with growing international reach. https://www.farsoon-gl.com

About LEAP 71
LEAP 71, strategically based in Dubai, UAE, is a pioneer in Computational Engineering. Its core technology, Noyron, is a Large Computational Engineering Model that autonomously designs physical objects by encoding logic, physics, and production methodologies. The company focuses on accelerating engineering across aerospace, electric mobility, robotics, and thermal systems, with a key goal of extending humanity’s footprint in space. https://leap71.com