The era of traditional search isn't ending with a whimper, but with a sharp pivot toward execution. Genspark, which started as just another AI-driven information synthesis engine, has effectively signed the death warrant for the "query-response" model by abandoning it entirely. In April 2025, the team unveiled Super Agent—a no-code platform where the final outcome matters more than polished prose. Market reaction was instantaneous: within 45 days of launch, a team of just 20 people reached an annual recurring revenue (ARR) of $36 million. They achieved this without bloated marketing budgets, relying solely on the shift from "summarize this for me" to "do this for me."
Architecture of Instant Utility
The technological backbone of this surge is a combination of GPT-4.1 and the OpenAI Realtime API. Unlike early AI assistants that functioned like basic chatbots with clunky plugin workarounds, Super Agent operates as a fully autonomous system. Under the hood, it orchestrates nine specialized LLMs and over 80 integrated tools, with GPT-4.1 serving as the primary engine for deep search and data analysis.
"OpenAI has supported Genspark from the beginning. Their APIs didn't just power our models—they allowed a 20-person team to build, launch, and scale faster than anyone thought possible," says Kay Zhu, CTO and co-founder of Genspark.
This speed to market is becoming the new gold standard for AI-native startups. The system handles everything from generating pitch decks to producing short-form video, sparing users from technical hurdles through a clean no-code interface.
Automating Human Nuance
The most disruptive—in the best sense—element of the Genspark stack is the Call For Me feature. Leveraging the speech-to-speech capabilities of the Realtime API, the agent conducts live dialogues with startling naturalism. The AI has learned to wait out hold music, respond appropriately to ambiguous answers, and navigate complex social dynamics. In Japan, the tool has already gone viral for a specific and high-stress use case: making resignation calls on behalf of employees.
Super Agent utilizes OpenAI’s multimodal models and Realtime API to automate workflows across text, images, and voice. Complex action chains are triggered by simple prompts without a single line of code.
This capability strikes at the heart of the traditional outsourcing and call center industries. When a 20-person startup provides a tool capable of real-time voice negotiations and complex content generation without developers, the value of custom communication systems and massive operations departments evaporates. The transition from search-based AI to action-oriented AI means the digital twin is no longer a sci-fi concept. It is a commercial reality capable of doing the heavy lifting in the real world while you focus on strategy—or simply take a break. Genspark proved it with the numbers: $36 million ARR in six weeks without a cent spent on paid ads is more than a success story; it’s a verdict on old ways of doing business.