The Internet has been a monumental force in shaping modern society, revolutionizing how we communicate, access information, and conduct business. It has broken down geographical barriers and language differences, connecting people and companies from all corners of the globe.

Over the years, the web has been meticulously optimized for human browsing. Search engines have been created to help people find what they’re looking for; user interfaces have been built to captivate visitors, encourage exploration, and facilitate desired actions; and a business model centered around advertising has been developed to capitalize on the most valuable resource for brands: human attention.

However, as artificial intelligence (AI) becomes increasingly powerful at automating tasks traditionally performed by humans, we’re on the brink of a new era — one where AI agents, rather than humans, are responsible for most of the web traffic.

But the web was not built for AIs to navigate it. AIs currently rely on systems that attempt to replicate how humans browse and interact with the web, something inefficient and ethically questionable as AIs thrive on structured data and don’t have attention to monetize.

At Linkup, we believe this new paradigm calls for the construction of new frameworks for search, interaction, and monetization, tailored for AIs.

We’re dedicated to participating in building them.

Toward AI-Optimized Search

More than a network of connected computers, the Internet has evolved into the largest repository of knowledge and services: more than 5 billion people use it in their daily life to access one of the 194 million active websites it hosts.

To allow us to navigate and access this immense repository, search engines have been developed, effectively becoming our portals to the web. And search is immensely complex: there are approximately 10 billion queries processed by search engines every day, requiring them to decide which of the 400 billion documents included in their Index to show you.

If Google commands a 90% market share in the space, it is because it provides the best ranking results: something it does by relying heavily on algorithms that are optimized for human users — prioritizing pages with high user engagement, user-friendly layouts, visual appeal, fresh content, and relevance to individual user factors like location and device.

For AI systems seeking information on our behalf, this human-centric optimization is deeply inefficient. AIs need rankings that go beyond keywords and metadata and that index and rank results based on a deeper understanding of context — search that leverages semantic relationships between words, concepts, and ideas. AIs can understand user intent like never before, and could make use of specialist search engines — tailored to specific domains and verticals — rather than the generalists that human-optimized search engines are today.

Yet, by lack of better alternatives, most AI systems today mimic human search behavior: typing queries into search engines and relying on rankings designed for humans. As this happens, AIs waste computational resources parsing through dozens of SEO-optimized websites and promoted links (often only scraping result snippets, which creates its own set of problems) while potentially missing crucial insights overlooked by keyword-based rankings.

We believe there’s a better way.

A New Framework for Interaction

Search is just the first step in web browsing. The true value of the Internet lies in its ability to allow humans to interact with content, people and companies. Reading a webpage to collect information is the most basic form of interaction, a significant part of the web’s activity involves much more complex actions, like publishing content, using software, or purchasing products and services.

Enabling interaction is far from trivial. To engage humans effectively, the web has evolved into a tapestry of visually rich designs, interactive elements, and intuitive navigation structures. UI and UX designers have meticulously crafted these elements to guide users seamlessly through content and websites. Efficient interaction is so central to digital products that studies have shown that companies prioritizing design outperform others by 219% on the S&P 500 over ten years.

But AIs don’t care about design. AIs don’t care about how webpages look or about how buttons animate when you click on them. AIs care about clean, fast, and reliable access to machine-readable data to function optimally.

Unfortunately, as with search, AIs lack tools to interact with content in a clean and efficient way and currently resort to replicating human behaviors. Developers work on incredibly complex systems, such as Visual language models, that try to understand and navigate through websites to make sense of how humans interact with the web in an effort to teach AIs to replicate such flows. Even for the most basic use case, retrieving information from a webpage, the only available option for AI systems is to use scraping: an ethically questionable, expensive, and error-prone solution. Visual elements that enhance human engagement and design systems that enhance human navigation represent noise and complexity for scrapers. In addition to causing inaccuracies, scraping is extremely computationally expensive, which is increasingly problematic given the need to minimize AI’s environmental impact.

We believe there’s a better way.

Rethinking the Internet’s Economic Dynamics

On average, people spend 6.5 hours online every day. For companies, every second is an opportunity to promote their products through advertising. As a result, the Internet’s architecture and business models have been intricately designed to capitalize on this time spent. This influences how content is created, presented, and consumed online, with many companies providing free services thanks to the monetization of human attention - Google and its search engine being the most prominent. On the internet, human attention is the product being sold to advertisers.

But AIs don’t have attention to sell. The rise of AI browsing puts pressure on the entire web architecture, with some arguing that AI could even “break the web.” We see this impact in several ways:

• Declining Advertising Revenues: Free content and services, including search engines, are monetized through ads that robots don’t consume when they scrap content. As AI traffic replaces human traffic, advertising revenues decline, threatening businesses that rely on these revenues, such as media companies. Some bilateral deals have started happening between big AI companies and publishers (parallel to the lawsuits), but these cannot scale for the thousands of smaller AI applications and media companies that lack the resources to negotiate.

• Uncompensated Use of Premium Content: Premium content is monetized through subscriptions and payments, reflecting its value. While AIs could theoretically pay for content, they differ from humans in that they train on the content they ingest. Without proper guardrails and agreements, premium content owners risk having their valuable content used without fair compensation. Some tools that are clicks away even specialize in scraping and reselling premium content to AI companies - in the absence of scalable alternatives, their proliferation could have dramatic impacts on the production of quality content creation, which is - paradoxically - a necessity for AI to thrive.

• Disruption of Customer Acquisition: Advertising helps businesses find customers. If human traffic decreases, companies must find new ways to reach their audience. AI browsing threatens the traditional customer acquisition funnel, especially for companies relying on existing search engines. As answer engines are more discriminating with the sources selected to draft their answers, data and content providers will likely concentrate around the companies most able to sign deals with these engines.

As a testimony of websites' fear of seeing their content stolen by AIs, websites are increasingly closing.

We believe this does not have to be a fatality — we believe new economic dynamics can flourish, centered around the monetization of direct API-based access, allowing content owners to regain control over how their material is accessed and monetized. Such solutions can promote new advertising dynamics, offering novel channels for companies to share their products with the world, leveraging AIs’ unique ability to precisely understand user intents.

An Internet for AIs

The World Wide Web is “an information system that enables content sharing over the Internet through user-friendly ways meant to appeal to users beyond IT specialists and hobbyists.” With the advancement of AI and its growing ability to automate tasks done by humans on the web, we believe that new frameworks need to be developed to enable content sharing over the Internet through AI-friendly ways.

At Linkup, our mission is to participate in building this new web, optimized for AI browsing. We’re dedicated to making this vision a reality through:

Step 1: Reimagining Search

We aim to rebuild search to be truly efficient at finding the most relevant information for AIs. No more sifting through dozens of websites with SEO-optimized but redundant content. Search should prioritize semantic relevance and data precision, tailored to the needs of AI agents.

Step 2: Rethinking Interaction

We need to make AI interaction with web content clean, fast, reliable, and transparent. By leveraging APIs and structured data formats, we need to allow AI agents to access and interact with content efficiently, while providing transparency and fairness to content owners. This isn’t just about accessing information; it’s about enabling agents to perform complex actions on our behalf over the Internet, requiring new infrastructures to do so effectively.

Step 3: Developing New Business Models

We must develop business models that ensure the right incentives and infrastructures exist for creators to continue producing quality knowledge. This means content owners should be fairly compensated for the use of their work by AIs while benefiting from transparency and traceability. These new models should be designed to allow companies and people to find their target audience even as human traffic is increasingly captured by AI platforms. Licensing agreements, API monetization, revenue sharing, and microtransactions are avenues to explore in creating an ecosystem that fairly distributes value across the board.

This new infrastructure will allow AIs to navigate the web 10x more efficiently and fairly than when it is shoehorned into using infrastructure designed for humans. This will allow a new set of super exciting applications to emerge — AI agents that automate tasks previously done by humans, liberating precious time to focus on higher value tasks that only humans can do. Such infrastructure should be focused on providing fairness and transparency to how content is used by AI systems, and ensure the value is fairly distributed to content creators, who will be empowered to invest more in research, innovation, and production value, knowing that their work can reach audiences more effectively through AI solutions.

This is the mission we’re passionate about at Linkup. If this vision resonates with you, we’d love to connect.