In the rapidly evolving landscape of 2026, the digital advertising industry has undergone a seismic shift from the "Chatbot-era" of simple assistants to the era of "Execution Agents." For growth teams at AI startups, the challenge is no longer just about writing a good prompt; it is about building autonomous loops that can reason through complex datasets and execute real-time changes across global platforms. This transition marks the end of manual dashboard management and the beginning of the OpenClaw revolution. This article provides a deep dive into the methodology used by Fruityo, an AI video platform, to scale their Google Ads spend while maintaining a consistent 1.8x ROAS through a multi-agent system nicknamed 'Agent Larry.'
The Shift from Chatbots to Execution Agents
For years, marketing automation meant setting up basic rules or using scripts that required constant supervision. Today, platforms like OpenClaw are redefining what is possible by utilizing the Model Context Protocol (MCP) to connect directly to Google Ads accounts, internal files, and messaging apps. Unlike traditional SaaS tools that live in a closed ecosystem, OpenClaw is a self-hosted AI agent that runs on your own infrastructure, providing unmatched security and flexibility. Growth teams are increasingly moving toward these agentic workflows because they report saving an average of 12 hours per week per client on manual auditing and reporting tasks.
The performance data backs this up. According to research from Stormy AI, ad copy optimized by OpenClaw agents currently sees a 2.5–4% conversion rate, which is nearly double the industry average of 1.2% for standard automated copy. This efficiency is why the OpenClaw project has exploded in popularity, surpassing 219,000 GitHub stars as of early 2026, making it the fastest-growing open-source marketing infrastructure in history. For developers and marketers alike, the GitHub repository has become the foundational text for modern growth engineering.
"The transition from SaaS to Agent-as-a-Service marks the point where software stops being a tool and starts being a teammate." — Peter Steinberger, Creator of OpenClaw.Case Study: How Fruityo Scaled to 1.8x ROAS with "Agent Larry"
The story of Fruityo, an AI-powered video platform, serves as the ultimate proof of concept for scaling ROAS with AI agents. The founder needed a way to manage a 24/7 marketing loop while traveling, without hiring a massive agency. They deployed "Larry," a custom-configured OpenClaw agent designed to manage their entire funnel. Larry wasn't just a script; he was a multi-agent team capable of coordinating organic reach on TikTok with paid scaling on Google Ads.
By using the AdWhiz skills directory, the team empowered Larry to monitor TikTok trends in real-time. When a specific UGC video started gaining traction—eventually reaching 5 million views—Larry didn't just report it; he automatically drafted Google Ads headlines based on the top-performing comments and visual hooks from the TikTok video. Larry then scaled the Google Ads budget to capitalize on the organic search lift, maintaining a 1.8x ROAS entirely through commands sent via Telegram. Much like how brands use Stormy AI to automate creator discovery, Larry automated the bridge between social trends and paid search execution.
Architecting the Multi-Agent Marketing Team
The secret to Fruityo’s success wasn't a single