Minitap’s $4.1 million seed round is the kind of velocity story that tends to define a cycle: two 23-year-old founders from rural France beat Google DeepMind on a core benchmark for mobile AI and, within months, convince a roster of unicorn operators and AI infrastructure insiders that they can reset the speed of mobile development itself. For retail and consumer app leaders whose roadmaps are increasingly constrained by mobile engineering bandwidth, Minitap is positioning its platform as a way to turn weeks of app work into days—and eventually, into an autonomous growth engine that iterates without human touch.
Why mobile speed now matters for retail
Mobile still dominates consumer attention—roughly 60% of internet usage—but app teams move far slower than their web counterparts, which is increasingly at odds with the rapid test-and-learn culture in modern retail and DTC. Internal data across consumer apps often shows a simple pattern: growth and product teams ship dramatically more experiments on web than on mobile, despite most revenue and frequency coming from mobile touchpoints.
The gap is structural more than strategic. Tooling for AI-assisted development has matured on the web (Cursor, Claude and others) but remains brittle on mobile, where code must run across devices, OS versions, and edge cases that are difficult for agents to observe and test in real time. For retail CMOs, heads of product, and growth leaders, that translates into slower pricing tests, fewer merchandising experiments, and delayed UX optimizations on the channel that matters most.
Inside Minitap’s technical stack
Minitap’s core bet is that agentic AI can be made reliable on phones, not just browsers, if you give it both fine-grained control of devices and a cloud-scale testbed. The company’s technical stack revolves around two pieces of infrastructure:
-
mobile-use, an open-source framework that lets AI agents control phones like human users—tapping, scrolling, and navigating across real Android and iOS apps.
-
minitap cloud, infrastructure that can spin up any phone configuration across thousands of devices in parallel, providing a realistic environment to run tests, catch regressions, and validate features across OS versions and hardware profiles.
These pieces connect into AI coding environments so agents can write mobile code, run it on real devices, identify bugs, fix them, and ship working features autonomously—all within a tight loop that aims to compress a traditional six-week mobile release into a matter of days.
The founders’ edge: AI research meets mobile craft
Nico and Luc’s story matters here because their credibility with investors rests less on pedigree and more on evidence that they can move faster than large research organizations on a hard technical frontier. Within their first 40 days, they claimed the #1 position on Google DeepMind’s AndroidWorld benchmark—an industry standard for measuring AI control of mobile devices—surpassing teams from Google DeepMind, Microsoft, ByteDance, and Alibaba.
That benchmark is not an abstract leaderboard. AndroidWorld is a dynamic environment built by Google Research to evaluate how well agents handle real Android apps across dozens of task types, from productivity and communication to content consumption. Beating sophisticated baselines on that suite suggests Minitap’s agents can generalize to the sort of fragmented, messy environments that define retail and commerce apps.
Investor lineup and what it signals
The $4.1 million seed round is co-led by Moxxie Ventures and Mercuri, both early-stage firms with deep networks in consumer and media-centric businesses, and joined by EWOR, Tekton Ventures, Amigos Venture Capital, and six unicorn founders. Backers include Thomas Wolf (Hugging Face), founders behind Last.fm, Adjust, SumUp, FlixBus, and operators from OpenAI, DeepMind, LangChain, and LlamaIndex—an investor group that blends AI research depth with experience scaling B2C platforms.
For retail leaders, that cap table is relevant because it indicates where Minitap is likely to land first: consumer mobile-first categories where rapid experimentation drives LTV—think language learning, wellness, marketplaces, and shopping apps—before moving into broader enterprise mobile footprints. Several early design partners reportedly came directly from customer founders who chose to invest after seeing the product, a pattern often associated with strong product–market fit in technical infrastructure.
Today, Minitap pitches itself as a way for engineering teams at consumer mobile companies to build features 10x faster, but the roadmap is explicitly aimed at empowering non-engineering stakeholders. The vision: a product manager or growth leader describes a feature, attaches a Figma design, and the platform generates code, tests it across devices, and ships an A/B test in a single afternoon—no tickets, no sprint planning, no handoffs.
