The Agentic Web: Building the Future with FIMO.ai

The internet, once built by humans for humans, is on the cusp of a profound transformation. Aurélien, co-founder of Strapi and now FIMO.ai, believes the web will soon be built by AI agents, primarily for other agents and humans. This shift, he argues, necessitates a new generation of AI-native tools, a gap he aims to fill with FIMO.

From Student Project to Global CMS: The Strapi Journey

Aurélien's journey into content management began as a student project in 2013. Alongside his co-founder, he started building websites as freelancers. The rise of mobile apps presented a challenge: managing the same content across both websites and mobile applications became cumbersome with existing tools like WordPress. This led to the creation of Strapi, initially named Wisti, as an internal solution for their clients.

After publishing Strapi on GitHub, they saw modest traction. Upon graduating, the trio decided to formalize their venture, founding a company with Aurélien leading product development as CPO, and his co-founder handling sales and company management. Today, Strapi is a fully remote company of 70 people, powering content for major global brands like Apple, Adidas, and IKEA. They have raised $50 million and boast millions of users worldwide.

Why Launch FIMO Now? The Cargo Ship Problem and the AI Imperative

Despite Strapi's success, Aurélien and his team faced a critical decision. Strapi was generating millions in revenue, nearing profitability, and had secured significant VC funding. However, the rapid advancements in AI were redefining the industry. They saw two paths: either witness an AI-native CMS emerge as a competitor, or build their own.

"It's not easy like when you have a product that 10 years old to it's like a big cargo ship like it's not easy to to change direction," Aurélien explains. Integrating AI deeply into Strapi, a mature product with long-term contracts and millions of users, would be a slow process. Recognizing that the "market windows is right now," they opted to build FIMO from scratch. This approach allows for rapid iteration, learning, and the application of those learnings back to Strapi, while mitigating the risk of disrupting their established product.

The Agentic Web: Who Builds the Internet Next?

FIMO's core thesis revolves around the concept of the "agentic web." Aurélien posits that the web, historically built and consumed by humans, is shifting. AI coding tools and agents are increasingly capable of developing and maintaining websites autonomously.

"The web was mostly built uh by humans for humans," he states. "The web I think is going to be built by agents mostly for agents and humans. And I think that's the big shift." This means AI agents will not only code websites but also manage their evolution, from updating content and responding to analytics drops to implementing defensive strategies against competitors. FIMO aims to empower users to build this first generation of "autonomous, proactive, and intelligent websites."

When Content Management Breaks: Lessons from Adidas

The challenges of scaling content management became evident through Strapi's work with large clients. Aurélien highlights the stark contrast between managing a 10-page website and a behemoth like adidas.com, which likely has tens or hundreds of thousands of pages.

"If you have a website with 10 pages, it's super easy to maintain all of them. But if you have like a giant website... It's super hard to maintain those pages to make sure like you don't have information that they're saying the opposite or making sure that the new tunnel voices apply everywhere on every pages." Even with a dedicated team, maintaining consistency and accuracy across vast amounts of content is a monumental task. This is precisely where AI's ability to manipulate text at scale can fill the gap, identifying and rectifying inconsistencies or informing human editors of potential issues.

The Human Limit: The Apple Christmas Thought Experiment

The increasing autonomy of AI raises questions about the boundaries of human control. Aurélien poses a thought experiment: what if an AI, tasked with maximizing revenue during the Christmas season, decided to remove all products from Apple.com except for AirPods, knowing they yield the highest margins?

"Is it a bad decision?" he asks. "If the goal of Apple at that time and the company is just to increase the revenue and maximize the profit and this agent is making the decision based on a lot of data points. Maybe Z end end up like um meeting their goals thanks to that decision made by an agent." While this scenario might feel unsettling, it underscores the potential for AI to make decisions that, while perhaps counterintuitive to human sensibilities, could be highly effective based on data. He believes that for many websites, especially those focused on SaaS or less brand-sensitive industries, AI can handle the bulk of content management, freeing humans to focus on creativity and strategic decision-making.

Workflows Nobody Actually Uses: The Strapi Learning Curve

Aurélien shares a surprising insight gleaned from Strapi's extensive user base: the perceived need for complex review and approval workflows is often overstated. While large enterprises like Adidas might require sophisticated systems for content produced 18 months in advance, the majority of users, particularly small to medium-sized businesses, do not leverage these features.

"Most of the people they think they need uh reviews and workflows but actually they don't," he observes. "They need that to reassure themselves. They need that to check a checkbox." For many, these workflows are more about perceived completeness than actual utility. Strapi found that less than 1% of its entire user base actively used these features, highlighting a disconnect between market demand and actual usage.

The Signal That Started It All: "I'll Just Recode That"

A pivotal moment for Aurélien came when Strapi customers, during enterprise sales negotiations, began stating, "I'm going to just recode that feature." This indicated that the value proposition of certain CMS features was diminishing as AI coding agents became more capable.

"The other thing that we miss and and I keep repeating myself on this is the maintenance and the maintenance is also cheap right now," he explains. If a developer leaves a company, an AI agent can now maintain that feature, removing the knowledge dependency that previously existed. This realization spurred the development of FIMO, an AI-native CMS designed from the ground up to embrace AI's capabilities, rather than retrofitting them onto an existing product.

Vibe Coding Is Eating the Market: The Threat to Traditional Builders

Aurélien sees the "vibe coding" category, powered by AI, as a disruptive force poised to reshape the entire website building market. Traditional players like WordPress, Wix, Squarespace, and even visual builders like Webflow and Framer, are all facing potential disruption.

"Do you really need like a visual editor if you can constrain the AI to use your brand?" he questions. "Do you need like a CMS if you actually don't really need workflows...?" The ease of prompting AI to generate unique, personalized websites offers a compelling alternative to template-based builders. This "IKEA effect," where users feel a sense of ownership and reward from building something themselves, is a powerful draw. He believes this consumer mass-market opportunity could even surpass the existing market size.

FIMO vs. Lovable: The Real Difference in AI Builders

The AI builder space is crowded, with many promising rapid website creation. However, FIMO differentiates itself by focusing on a structured content layer that AI can truly reason about and manipulate.

Aurélien contrasts FIMO with tools like Lovable, which he acknowledges are excellent for building applications but less suited for content-centric websites. Lovable, he notes, often generates static content embedded within the code, lacks robust SEO optimization, and provides a raw, developer-focused interface for content editing.

FIMO, on the other hand, employs three specialized agents: one for design, one for coding, and crucially, one for content management. This ensures content is stored in a proper CMS with a user-friendly interface, rich text editing, and a media library. "Because we have made the the content is structured, we can play with it," Aurélien explains. This structured content becomes a contextual asset for AI agents, enabling them to compare, translate, and adapt content across an entire website, a capability largely absent in other AI builders.

What Makes FIMO Actually Different: Structured Content and Autonomous Websites

The core innovation of FIMO lies in its structured content approach, managed by dedicated AI agents. This allows for a level of manipulation and autonomy previously unattainable.

"The content created is not exactly the same. it will better fit your brand industry and and goals and we always store the content in the CMS," Aurélien states. This means users get a real CMS interface, complete with rich text editing and media management. Crucially, this structured content serves as a rich context for AI agents. If a competitor updates their pricing, FIMO's agents can analyze and react to the user's entire content base, not just isolated snippets of code.

Furthermore, FIMO is building towards fully autonomous websites. Agents can monitor competitors, convert Slack discussions into blog posts, or automatically generate content based on GitHub repository updates. The goal is to align AI actions with specific business objectives, whether it's optimizing sign-ups, demo requests, or lead generation.

What Does AI-Native Really Mean? Concrete Principles at FIMO

Being "AI-native" goes beyond simply integrating AI features. For FIMO, it means fundamentally rethinking product design and development processes.

"We never start start from a blank page," Aurélien reveals. Instead, FIMO prompts users for their desired outcome, generating a first version that can then be iterated upon. This avoids the common user experience of facing an empty canvas.

The technical stack is also entirely different. "90% of the code is being AI coded," a stark contrast to Strapi's development. The team also embraces a "vibe-coded QA" approach, where engineers and designers directly fix issues as they encounter them, drastically reducing the time from identification to resolution – from weeks to minutes. This agile, AI-assisted workflow allows for rapid feature development, enabling FIMO to rebuild many Strapi features with a small team in just six months.

The PM Role Is Being Automated: Focusing on Vision and Strategy

The rise of AI is automating many of the routine tasks previously handled by Product Managers (PMs). Aurélien believes this is a positive development, freeing PMs to focus on higher-level strategic work.

"I don't like to see PMS spending their time like doing task that can be done by AI where your real value is about making the right decisions," he asserts. Tasks like writing transcripts, triaging feedback, writing user stories, and managing backlogs can now be handled by AI. This allows PMs to dedicate more time to understanding the market, testing new products, developing their "product sense," and making informed decisions.

The Skills That Matter Now: Product Sense and Market Feel

For PMs navigating the AI era, the emphasis shifts from execution to strategy. Aurélien stresses the importance of a clear vision, deep market understanding, and strong "product sense."

"You need to control where your product is is going without controlling it," he explains. This means setting a clear direction and trusting the team, empowered by AI tools, to execute. He advises PMs to embrace AI, automate mundane tasks, and spend more time observing competitors, understanding market trends, and testing new technologies. The ability to guide engineers towards the right direction, ensuring their proactive efforts are aligned with customer needs, remains a critical PM responsibility.

What FIMO Learned After Launch: Zero Churn and Unexpected Users

Since launching FIMO, Aurélien and his team have gained valuable real-time insights. One surprise was the initial customer base: a significant portion of their early users are in China, accessing FIMO via VPN. This was unexpected, as they had anticipated a more European user base.

Another key learning is that their customers are not coders. While they may not be afraid of seeing code, they don't actively code. This validates FIMO's approach of abstracting away the complexities of coding. Perhaps most remarkably, FIMO has experienced zero churn. This contrasts with the high churn rates seen in other AI tools, suggesting that FIMO's focus on structured content and a clear value proposition creates a stronger hook.

Advice for Founders & PMs: Embrace AI, Small Teams, and No Constraints

For founders building AI-native products, Aurélien's advice is clear: "Do not optimize the economics right now." Focus on user adoption and usage, even if it means operating with zero or negative margins initially. He also emphasizes the power of small, lean teams for innovation. Creating FIMO involved a compact team working in isolation, free from the constraints of supporting a larger, established product.

For PMs working within existing non-AI-native products, the challenge is to advocate for change. They must identify market windows and present compelling arguments for either adapting the existing product or starting anew with a focused, AI-native offering.

What Keeps Aurélien Up at Night: A Web You Can't Trust

Looking ahead, Aurélien expresses both excitement and apprehension about the future of the web. He foresees a web filled with AI-generated content, raising questions about trust and authenticity.

"Can we trust what we are seeing on the screen. And I think the the answer is no. you can't trust it anymore," he admits. This shift, he believes, will elevate the importance of real-life experiences and human creativity. While he's committed to building the "AI way" for the agentic web, he acknowledges the profound societal implications and trusts that industry leaders will navigate these challenges.

Rapid Fire Questions

Aurélien's insights offer a grounded perspective on the AI revolution, emphasizing practical application, strategic foresight, and the fundamental shift towards an agentic web.