Tomasz Woźniak
Supervisory Board Member
If we take a closer look at Meta’s revenue structure, one thing becomes immediately clear: even the most hyper-personalized advertising is still advertising. For years, other Big Tech companies have been actively diversifying their revenue streams: cloud, hardware, subscriptions, fintech, marketplaces. Meta has been moving noticeably slower in this area.
If no decisive move is made, an ad‑centric business model based primarily on monetizing a massive user base will eventually lose relevance. Users are already experimenting with alternative communication tools, AI‑native interfaces, and new digital habits that weaken the traditional social media value loop.
To understand the superapp model, it is enough to look at a single example: WeChat. The world’s most successful superapp was built directly on top of a communicator. Messaging was not an add‑on but the foundation.
WeChat became a daily operating system for users because conversation naturally sits at the center of digital life. From chat, users moved seamlessly into payments, services, bookings, content, and commerce. No complex navigation, no context switching, just conversation turning into action.
Messaging is the most powerful interface for orchestrating complex digital behaviors at scale.
Outside of Asia, no company has truly replicated this model at global scale. Today, Meta is uniquely positioned to do so. With WhatsApp, Instagram, and Messenger, it owns the world’s most widely used conversational interfaces, and it has already experimented with broadening the scope of these services, like with WhatsApp Pay, which allows users in India and Brazil to easily exchange money in between messages.
But copying WeChat one‑to‑one is not the point. The opportunity looks different today. Instead of mini‑programs and payments alone, the next generation of superapps will be powered by dedicated, task‑specific AI agents that can book, plan, recommend, negotiate, create, and execute on behalf of users.
This is where the interface matters as much as the intelligence. Agents without distribution remain demos. Distribution without intelligence becomes a commodity.
There is, however, an important constraint in Meta’s path forward. Apple and Google both have access to the same users, often one step earlier, at the operating system level. This effectively positions them as potential gatekeepers of any consumer‑facing AI experience.
Apple, in particular, has already demonstrated its power in this role. With the introduction of App Tracking Transparency in iOS 14.5, Apple significantly limited cross‑app profiling. The result was not only a structural blow to ad‑based business models, but also a behavioral reality check: only around 20–25% of users globally opted in to tracking.
This episode made one thing clear: control over the OS layer translates directly into control over data access, monetization mechanics, and ultimately the shape of consumer experiences.
In this context, the acquisition of Manus can be seen as a strategic leapfrog toward the end consumer. If the deal goes through, Meta would be accelerating directly into AI‑native consumer workflows rather than building incrementally from existing ad products.
At the same time, Meta is fighting a parallel war at the model layer, with OpenAI and other frontier AI players pushing aggressively on intelligence, reasoning, and multimodal capabilities. Competing only on models is capital‑intensive and increasingly commoditized.
Owning both the intelligence layer and the interface layer may be the only sustainable advantage. Manus fits precisely into this strategy: shortening the distance between advanced AI capabilities and everyday consumer use.
This dynamic also explains why Meta continues to experiment with hardware, from smart glasses to other wearables. Hardware creates a parallel distribution layer that bypasses traditional OS‑level gatekeeping. It also unlocks a more efficient path to Edge AI acceleration, where on‑device intelligence can operate with lower latency, better privacy characteristics, and tighter integration into everyday behavior.
In a world where AI agents increasingly live closer to the user, hardware becomes not a side project, but a strategic necessity. Miniaturization is advancing at a dizzying pace and the progress is evident.
For Meta, this is no longer about advertising optimization. It is about redefining how consumers interact with digital services altogether.
Seen through this lens, Meta’s strategy is less an isolated bet and more an early signal of where the superapp race is heading globally. Almost every serious platform player is now probing the same direction from a different angle: Apple through OS-level intelligence and private, on-device agents; Google through search-to-task execution and Android-native assistants; Amazon through commerce-embedded agents; and Chinese platforms by deepening already integrated ecosystems.
What differentiates Meta is not superior intelligence, but control over conversational distribution at planetary scale. That position allows it to test AI-native execution layers directly inside daily communication flows, long before most competitors can meaningfully follow.
For large-scale consumer platforms such as Allegro, InPost, or retailer ecosystems like Żabka, the lesson is not to wait for a dominant interface to emerge, but to prepare for a world where interfaces fragment while execution converges.
The strategic focus should remain on building defensible, structured data foundations: a high-fidelity single customer view that supports both precise segmentation and deep behavioral modeling. With user bases exceeding 10 million, many of these platforms already have sufficient scale to train domain-specific models and evolve toward dedicated AI agents that act on behalf of customers across discovery, decision, and fulfillment.
Crucially, the final channel of reach will be increasingly agnostic. Some customer segments will prefer WhatsApp, Messenger, or other conversational environments to execute brand-related tasks, while others will remain inside first-party applications. The winning platforms will not attempt to force users into a single interface.
Instead, they will decouple intelligence and execution from presentation, allowing their agents to operate consistently across multiple channels while remaining anchored in proprietary data, trust, and brand relationship. In that sense, the next generation of superapps will not look like apps at all. They will function as portable execution layers, embedded wherever user intent happens to surface.
P. Wasilewski - Head of Solutions
J. Dogadalski - Business Development Lead
J. Dogadalski - Business Development Lead
M. Klimczak - VP of Software Delivery
