Droidrun's answer
Droidrun was founded by a tight-knit team of four co-founders with a history of successful collaboration. “We’re four co-founders who’ve already built two companies together, which gives us a proven working dynamic,” said Christian.
Christian, launching his third company, focuses on strategy and identifying early market opportunities. Niels Schmidt, Chief Technical Officer, architected the core mobile automation technology. He has deep experience in AI and blockchain, “but more importantly, he builds things that work in production,” Christian noted.
Nikolai Dück, Chief Product Officer, built an AI chatbot platform seven years ago, long before AI entered the mainstream, that Fortune 50 companies still use today. Peter Lächner, who previously worked at Bosch managing international teams, brings crucial expertise in how large enterprises purchase and deploy technology. Before Droidrun, the team built a dominant local loyalty platform in a German city and bonny.network, a blockchain-based consumer data reward platform.
The momentum for Droidrun came from a shared frustration. “We needed AI agents to extract product insights from viral TikTok videos for our previous startup. Sounds simple, but every mobile automation solution failed catastrophically. UiPath, Selenium, academic tools: they’d work in demos but break the moment TikTok updated their interface,” Christian recalled. This pain revealed a fundamental missing piece in the AI automation industry: reliable access to mobile app interfaces.
Traditional AI mobile automation tries to “see” the screen like humans, capturing screenshots, analysing pixels, and estimating button locations. This approach is slow, fragile, and breaks when apps update.
“We flipped it completely. Instead of teaching AI to interpret mobile screens visually, we convert mobile interfaces into structured text that large language models already understand. Rather than hunting pixels, we give AI the actual button data and interface structure,” Christian explained.
This approach delivers three times faster execution, dramatically lower computational costs, and much higher reliability when apps update their UIs. Droidrun’s solution earned the #1 rating on the AndroidWorld benchmark and, crucially, works in production environments where vision-based solutions fail.
Droidrun combines visual understanding with UI structure extraction, enabling reliable control of mobile apps on Android and iOS. Its open-source framework has attracted rapid community adoption, with over 900 developers in 24 hours and more than 3,300 GitHub stars in under three months, making it the fastest-growing mobile AI agent project.
Looking ahead, Droidrun is accelerating the development of its cloud platform, enabling enterprises to deploy mobile AI agents at scale while maintaining the open-source framework that fuels developer innovation.
Christian outlined the vision for the coming years: “Within a few years, automated workflows will outnumber human interactions on mobile devices. Instead of switching between a dozen apps to book a business trip, checking flights on one, hotels on another, and expenses in a third, you’ll simply tell your AI agent, ‘Book my trip to Berlin next week for the client meeting.’ The agent will handle it all automatically.”
This future promises a profound transformation. “Instead of static app icons and manual navigation, we’ll have agentic user interfaces that adapt and respond intelligently, or entirely new operating systems designed around AI agents rather than human touch.”
He stressed the essential role of infrastructure: “For this transformation to happen, agents need seamless access to mobile systems to bridge human intent with digital action. Droidrun is building the infrastructure that makes this revolution possible.”
Early adopters are using Droidrun to unlock data from mobile-only platforms, automate critical business workflows, extend robotic process automation (RPA) to mobile devices, support QA automation, and build AI agents that require mobile execution.
BrowserStack
Evebyte Studio
Neura
And-GPT
Browser Use
Runner H
Skyvern
Sauce Labs
browserling
TestMu AI (Formerly LambdaTest)
Browsershots
Litmus
MultiBrowser
