• The Rise of GEO and the Agency Response

• What is Generative Engine Optimization?

• The Vibe-Coding Revolution: Building Tools in Hours

• Case Study: Havas and Brand Insights AI

• Beyond Havas: Broadhead and Supergood Join the Movement

• Why Build In-House? The Case for Control and Customization

• The Cost Calculus: Avoiding Enterprise Lock-In

• The Competitive Landscape: Startups vs. Agency-Built Solutions

• Technical Implementation: How Claude Code and Replit Power Innovation

• Global Reach: Scaling AI Tools Across Languages and Markets

• The Future of Agency-Client Relationships in the AI Era

• Challenges and Considerations for Agency-Built Tools

 

 

1. The Rise of GEO and the Agency Response

The advertising and marketing world is witnessing a seismic shift. As consumers increasingly turn to generative AI platforms like ChatGPT, Perplexity, and Claude to find information, make purchasing decisions, and discover brands, a new discipline has emerged: Generative Engine Optimization, or GEO. This practice, the natural successor to traditional search engine optimization, focuses on ensuring that brands appear prominently and positively in the answers generated by large language models. In response to this paradigm shift, a wave of startups has emerged, promising to help brands track and improve their visibility in AI responses. However, a fascinating counter-trend is taking root within the agency world. Creative powerhouses like Havas, Broadhead, and Supergood are bypassing third-party vendors altogether and building their own GEO tools internally. Leveraging cutting-edge coding assistants such as Anthropic's Claude Code, these agencies are "vibe-coding" bespoke applications in a matter of hours, tailoring them precisely to their workflows and client needs. This article explores this phenomenon, delving into the motivations, methods, and implications of agencies taking AI development into their own hands.

 

 

2. What is Generative Engine Optimization?

 

 

3. The Vibe-Coding Revolution: Building Tools in Hours

The term "vibe-coding" might sound whimsical, but it describes a very real and transformative shift in software development. It refers to the practice of using AI-powered coding assistants to build applications rapidly, often through natural language prompts and iterative refinement, rather than traditional, line-by-line programming. For ad agencies, this has been a game-changer. Teams that may not have had deep coding expertise can now, with the help of tools like Anthropic's Claude Code, create functional, sophisticated applications in a fraction of the time it would have taken just a year ago. The process is intuitive: a strategist or data analyst can describe the tool they need "an application that analyzes how our client's brand appears in ChatGPT responses" and the AI assistant helps generate the code, debug errors, and suggest improvements. What once required a dedicated development team and weeks of work can now be accomplished by a small, agile team in a matter of hours. This democratization of software development is empowering agencies to move with unprecedented speed, prototyping and deploying custom solutions that address their specific needs without the overhead of traditional software procurement.

 

 

4. Case Study: Havas and Brand Insights AI

A leading example of this trend is Havas, the global advertising and communications giant. The agency has developed Brand Insights AI, a proprietary GEO product built using Claude Code and the Replit development platform. The tool's functionality is elegantly simple yet powerfully effective. It generates a series of prompts based on a client's brand and its competitive landscape. These prompts are then run across multiple large language models simultaneously. Brand Insights AI analyzes the resulting responses, tracking how often and in what context the client's brand appears, including citations and attributions. In essence, it simulates how the brand shows up in AI-driven discovery, providing invaluable data for optimizing their GEO strategy. According to Dan Hagen, Havas's global chief data and technology officer, the platform has been rolled out globally, covering nearly 100 countries and more than 60 languages. It is not merely an internal tool; it is licensed to clients as a software-as-a-service product. Furthermore, it has become a core component of the agency's pitch strategy and has directly contributed to winning new business. Brand Insights AI is a testament to the power of in-house innovation, transforming a technological capability into a competitive differentiator and a new revenue stream.

 

 

5. Beyond Havas: Broadhead and Supergood Join the Movement

Havas is not alone in this endeavor. Other agencies, both large and small, are recognizing the strategic value of building their own GEO tools. Broadhead, a full-service marketing agency, and Supergood, a creative agency, are among those interviewed for this story who are actively developing in-house systems. For these agencies, the motivation is similar: a desire for tools that fit seamlessly into their existing workflows and can be customized to the unique needs of their clients. Off-the-shelf solutions, while useful, often come with rigid structures and generic features that don't align with an agency's specific processes or the nuanced requirements of different brands and industries. By building their own tools, agencies like Broadhead and Supergood can ensure that the technology serves them, rather than the other way around. They can integrate GEO analysis directly into their campaign planning, creative development, and reporting, creating a more holistic and effective approach to AI-driven marketing. This grassroots movement signals a broader trend: agencies are no longer just consumers of marketing technology; they are becoming creators of it.

 

 

6. Why Build In-House? The Case for Control and Customization

 

 

7. The Cost Calculus: Avoiding Enterprise Lock-In

Beyond the desire for control, there is a pragmatic financial consideration driving the build-versus-buy decision: cost. Hagen has revealed that Havas has so far opted against signing an exclusive enterprise agreement with a large language model provider like Anthropic. Such agreements, he notes, can run into "multiple millions" of dollars annually. For an agency with thousands of employees and varying levels of AI adoption, committing to such a massive expenditure is a risky proposition. "It's a combination of flexibility," Hagen explains. "It would be challenging for me to sign four or five enterprise agreements just from weight of cost." He also points to the challenge of "cost control and management." AI usage within an agency is not uniform; some employees are deeply embedded in AI workflows, while others are still learning. Committing to thousands of enterprise licenses risks paying for capacity that is not yet fully utilized. "We didn't want to be in a position where we're paying for ten thousand licenses that people are using once a week," he states. By building their own tools and managing their consumption of underlying AI models on a usage basis, agencies can align their costs with actual demand, avoiding the financial drag of unused enterprise licenses.

 

 

8. The Competitive Landscape: Startups vs. Agency-Built Solutions

The emergence of agency-built GEO tools places these creative firms in a fascinating position relative to the startup ecosystem. A wave of venture-backed startups, including Profound, Bluefish, and Emberos, has emerged specifically to help brands track and improve their visibility in AI responses. These companies offer specialized platforms and are betting that brands and agencies will outsource their GEO needs to experts. However, the agencies building their own tools are effectively choosing to compete, at least internally, with these startups. They are arguing that their intimate understanding of their clients' brands, their creative strategies, and their internal workflows gives them an advantage that a generic platform cannot match. This does not mean that startups are irrelevant; they may still find a market with smaller agencies or brands that lack the resources to build their own solutions. However, for larger, tech-savvy agencies, building in-house is a way to differentiate themselves, retain more value, and maintain control over their strategic destiny. It represents a form of vertical integration, where the agency becomes both the user and the creator of its core technological tools.

 

 

9. Technical Implementation: How Claude Code and Replit Power Innovation

The technical enablers of this agency-led revolution are the new generation of AI-powered development tools. Havas's Brand Insights AI was built using two key platforms: Anthropic's Claude Code and Replit. Claude Code is an AI assistant that can understand, generate, and debug code based on natural language prompts. It acts as a collaborative partner to human developers, accelerating the coding process and lowering the barrier to entry for those with less formal programming training. Replit is a browser-based integrated development environment that makes it easy to write, run, and deploy code collaboratively. The combination of these tools creates a powerful and agile development workflow. A small team can ideate a tool, use Claude Code to generate the initial codebase, refine it through iterative prompts, and deploy it via Replit, all in a fraction of the time of traditional methods. This technological stack is perfectly suited to the fast-paced, experimental culture of modern ad agencies, enabling them to prototype, test, and launch new capabilities with remarkable speed and efficiency.

 

 

10. Global Reach: Scaling AI Tools Across Languages and Markets

One of the most impressive aspects of the agency-built tool movement is its global scale. Havas's Brand Insights AI, for example, has been rolled out across nearly 100 countries and supports more than 60 languages. This is not a niche experiment; it is a core, enterprise-wide capability deployed across a vast, multinational organization. The ability to analyze brand presence in AI responses across dozens of languages and cultural contexts is a powerful asset for global brands. It allows them to understand not just how they appear in English-language AI models, but how they are perceived in French, Japanese, Arabic, and countless other linguistic markets. This global perspective is essential for maintaining brand consistency and tailoring messaging to local audiences in the age of AI discovery. The fact that an agency-built tool can achieve this level of scale is a testament to the sophistication of the underlying technology and the strategic vision of the agencies deploying it. It demonstrates that in-house development is not just for small, experimental projects; it can power core, global business functions.

 

 

11. The Future of Agency-Client Relationships in the AI Era

The development of proprietary AI tools by agencies has profound implications for the future of agency-client relationships. In the traditional model, agencies provided strategic and creative services, often relying on third-party technology platforms to execute and measure campaigns. By building their own tools, agencies are adding a new dimension to their value proposition. They are becoming technology partners, not just service providers. A tool like Brand Insights AI becomes a tangible asset that the agency can offer to clients, either as part of a broader engagement or as a standalone SaaS product. This deepens the client relationship, creates new revenue streams, and positions the agency as an indispensable partner in navigating the complexities of the AI era. Clients benefit from having a tool that is deeply integrated with the agency's strategic thinking and creative process, rather than a generic platform that requires separate onboarding and management. This trend points towards a future where the line between agency and technology company becomes increasingly blurred, with creative firms competing on the strength of both their ideas and their proprietary software.

 

 

12. Challenges and Considerations for Agency-Built Tools

While the benefits of building in-house GEO tools are compelling, this path is not without its challenges. Developing and maintaining software requires ongoing investment in talent, infrastructure, and security. Agencies must ensure that their tools are reliable, scalable, and protected against data breaches. They must also keep pace with the rapid evolution of the underlying AI models, which can change their behavior and output with little notice. A tool that works perfectly with one version of a model may need significant adjustment after an update. Furthermore, there is the strategic risk identified by Hagen himself: "What if they're not frontier enough in six months?" Committing too deeply to a single model provider could leave an agency stranded if a superior competitor emerges. This is why agencies like Havas are opting for flexibility, avoiding exclusive enterprise deals and building tools that can be adapted to work with multiple models. The decision to build in-house is not a one-time event; it is an ongoing commitment to innovation, adaptation, and strategic vigilance.