minitap
Founded 2025 · 14 employees on LinkedIn · 15 known investors
Minitap builds "mini" (minitest), an autonomous AI agent for mobile app QA that connects to a team's product specs, tickets, and codebase to map app flows and run end-to-end regression and PR tests across iOS, Android, and web. It produces self-healing tests with session replays, logs, and fix prompts, and integrates into CI pipelines and tools like Slack, serving mobile engineering teams.
Also known as mini · Minitap · minitap.ai · minitest
Investors · 15
Also in the syndicate · 11
Company profile
researched Aug 2026Minitap is a Paris-based company developing "mini," an AI agent positioned as an autonomous QA teammate for software teams. mini connects to a company's product specifications, tickets and codebase, learns how the product works, maps its user journeys, and determines what changes with each release. It then tests individual pull requests based on what is at risk, runs full regression suites, investigates failures, and returns reports to engineering teams. The company contrasts this with manual QA and script-based end-to-end frameworks such as Maestro, Appium, Playwright and Cypress, where humans design suites, author tests and maintain them; with mini, the agent handles suite design, authoring, change adaptation and failure triage, while humans retain the decision to ship.
The product, released under the name minitest, runs tests on iOS, Android and web, including virtual and real devices, and covers scenarios such as OAuth and OTP login, offline and flaky network conditions, file and media uploads, personas, GPS and location mocking, and runtime CPU/memory logging. Tests are expressed as outcome-based criteria derived from code, PRDs or a one-line description rather than brittle selectors, with the stated aim that suites survive refactors without authoring or maintenance. Failures are delivered with session replay, expected-versus-actual comparison, repro steps, device logs, and a fix prompt formatted for coding assistants such as Cursor and Claude. mini runs inside customer CI pipelines, returning a pass/fail check on the release pull request or a message in Slack, and integrates with GitHub, Slack, Gmail and Jira, with Linear, Bitbucket, GitLab and Notion listed as forthcoming. Supported mobile stacks include React Native, Flutter, Swift and Kotlin. Optional human QA engineers from Minitap can be added for review.
Minitap also maintains mobile-use, an open-source Python agent framework (Apache-2.0) that lets AI agents control real Android and iOS devices and simulators through natural language, including UI-aware navigation and structured data extraction, and is configurable across LLM providers including OpenAI, Google Vertex AI, Anthropic, xAI, OpenRouter and MiniMax. Related public repositories include a minitest CLI, a GitHub Action for triggering minitest suites, and a Homebrew tap.
Founding story
Minitap was founded in 2025 by Nicolas Dehandschoewercker (co-founder and CEO) and Luc Mahoux-Nakamura, both 23 at the time of the seed announcement. The two grew up together in Cosne-Cours-sur-Loire, a village in the Burgundy region of France, where Luc taught Nico to code; they studied together to rank in France's top 0.1% academically and have worked together on projects for about seven years. At 18 they built their first mobile app and bootstrapped a product called Fuego to 10,000 users. Nico studied biomedical engineering at Imperial College London and worked on AI research; Luc built delivery drone infrastructure at Rakuten. The CEO has said the company was started to shorten the distance between building a product and learning whether users want it, motivated by the two years spent building their first mobile product.
Business model
Minitap sells its mini/minitest agent to engineering teams as a service that replaces or supplements in-house QA work. The site presents plan tiers with a monthly estimate generated from a short questionnaire about the platforms and surfaces a team ships on, plus a free trial and a live demo. Minitap's own QA engineers can be added for human review as an optional layer on top of the agent. Deployment options span Minitap's cloud for smaller teams, a customer's VPC, and a customer's own datacenter or fully air-gapped network for the largest accounts, including use of physical devices in customer racks.
Subscription plans priced against the platforms and volume of release checks a customer requires, with a monthly estimate produced from the site's plan calculator; a free trial is offered and a promotional 50% first-year discount was offered to partners booking a demo around the minitest launch.
Traction
The company reports more than 20 customer companies and states that apps tested by mini reach over 100 million people. Memrise is cited as a long-running customer, alongside Nanogram and a stealth EdTech company. The open-source mobile-use repository grew from roughly 1,900 GitHub stars at the time of the December 2025 seed announcement to about 2,800 stars and 237 forks with 879 commits in the GitHub listing. Press coverage states early customers were using Minitap to accelerate experimentation cycles on consumer-facing mobile apps.
Latest developments
In May 2026 Minitap announced the release of minitest, offering partners who booked a demo a 50% discount for the following year, and expanded availability from mobile to web. Public repositories including the minitest CLI, a minitest-trigger GitHub Action and mobile-use show activity through August 2026. The company is hiring in Paris, including a full-time forward deployed engineer role advertised at €50k–€80k depending on experience, to own customer deployments in Minitap's cloud, customer VPCs, on-premise datacenters and air-gapped networks.
▸Full profile — market position, technology, go-to-market, geography, history, risks & controversies
Market position
Minitap positions mini against manual QA, script-based end-to-end automation frameworks (Maestro, Appium, Playwright, Cypress) and "agentic QA" tools that still require humans to author and maintain plain-English flows, claiming full agent ownership of the test lifecycle apart from the ship decision. On Google DeepMind's AndroidWorld benchmark for mobile-app agents, the company's open-source agent is listed first at a 100% success rate, ahead of entries from The AGI Company (97.4%), askui (94.8%), H Company's Surfer 2 (87.1%), gbox.ai (86.2%) and Z.AI (80.2%). Investors describe Minitap as among the first companies applying agentic AI to mobile development with a full-stack approach.
Minitap's stated differentiator is full agent ownership of the QA lifecycle: the agent reads the codebase and specs and requires zero test authoring or maintenance, as opposed to script-based frameworks or agentic tools where humans still write and maintain plain-English flows. Tests are defined as outcome criteria rather than selectors so they survive refactors, the agent asserts on rendered UI like a user and can validate AI-driven features that scripts cannot assert against, and every failure ships with replay, logs, repro steps and a ready-to-use fix prompt. The company also points to its AndroidWorld benchmark result and its open-source, framework-agnostic agent, plus deployment flexibility up to air-gapped customer datacenters.
Technology
Two core technologies underpin the product: mobile-use, an open-source framework that allows AI agents to control real Android and iOS devices and simulators using natural language and the accessibility tree, and minitap cloud, infrastructure that spins up thousands of iOS or Android configurations in parallel. The agent asserts on rendered UI rather than selectors, generates tests from code, PRDs or one-line criteria, self-heals as the app evolves, and captures runtime logs and session replays. mobile-use is written in Python, licensed Apache-2.0, ships via Docker with quick-launch scripts for Linux/macOS and Windows, and supports pluggable LLM providers (OpenAI, Google, Anthropic, xAI, OpenRouter, MiniMax and OpenAI-compatible or local endpoints).
Go-to-market
Direct sales supported by booked demos and a self-serve free trial and live demo at app.minitap.ai, amplified by open-source distribution of the mobile-use agent framework on GitHub and a Discord community. The company publicizes its AndroidWorld benchmark result and research paper as technical proof points, publishes video content about product launches and team life ("mini TV"), and is hiring forward-deployed engineers accountable for customer deployment and adoption outcomes.
Mobile engineering teams, particularly small to medium-sized teams at consumer app companies shipping frequently on iOS and Android, as well as teams shipping web surfaces. Named or referenced users include Memrise, Nanogram and an unnamed stealth EdTech company, with the site citing more than 20 companies. Enterprise customers with strict security policies requiring on-premise or air-gapped deployments are also addressed.
Geography
Headquartered in Paris, France, where the roughly 10-person team and open roles are based. The seed funding press release was datelined San Francisco, and the company states it will fly the whole team to San Francisco when needed.
History
Founded in 2025, the team reached the top position on Google DeepMind's AndroidWorld benchmark within 40 days of development and then open-sourced its mobile-use agent framework. Roughly four months later, on 1 December 2025, the company announced a $4.1 million seed round co-led by Moxxie Ventures and Mercuri. In May 2026 it announced the release of minitest, the productized agent that owns a team's mobile test suite end to end, and by that point described itself as a 10-person team in Paris hiring for roles including forward deployed engineer.
Risks & controversies
Publicly available material is largely company-authored, including a paid press release distribution of the funding announcement. Performance claims such as "10x faster mobile development," the 100M+ user reach figure and the customer count are company statements rather than independently verified data. The open-source mobile-use agent is documented as having limited effectiveness with games because they do not expose accessibility tree data, and its Docker quickstart supported only Android devices and emulators, with physical iOS device support constrained. The company also identifies concentration risk in that customer deployment and adoption responsibilities currently fall to the founders, which it describes as the biggest constraint on how many customers it can support.
Compiled by commissioned research from 8 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Competitors · 8
by search overlapCompanies competing with minitap for the same Google search keywords, organic and paid, via search-intersection analysis.
Timeline · 4
launches, deals, and filingsMinitap announced the release of minitest, in which the mini agent owns a mobile team's test suite end to end — authoring, execution, triage and fix prompts — running on real iOS and Android devices and delivering session replays, device logs and Cursor-ready fix prompts to Slack.
Minitap announced a $4.1 million seed round co-led by Moxxie Ventures and Mercuri, with participation from EWOR, Tekton Ventures, Amigos Venture Capital, six unicorn founders (including Thomas Wolf of Hugging Face, Petter Made of SumUp, Paul Muller of Adjust, Last.fm and FlixBus founders, and Saturnin Pugnet of Worldcoin), and operators from OpenAI, DeepMind, LangChain and LlamaIndex.
$4.1M source ↗
Minitap's open-source mobile-use agent reached the top global position on AndroidWorld, the benchmark for AI-controlled mobile devices, ahead of teams from Google DeepMind, ByteDance, Microsoft Research and Alibaba; the company reports a 100% success rate on 116/116 tasks across 20 apps, achieved within 40 days of development and roughly four months before the seed announcement.
After the AndroidWorld result, the founders open-sourced mobile-use, a framework that lets AI agents control Android and iOS devices with natural language; the repository is Apache-2.0 licensed and supports multiple LLM providers including OpenAI, Google, xAI, OpenRouter, Anthropic and MiniMax.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
▸Research sources · 8
primary sources listed
- minitapminitap.ai · web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does minitap do?
- Paris-based Minitap builds mini, an autonomous AI agent that owns end-to-end QA testing for mobile and web apps.
- Who are minitap's investors?
- minitap's investors include EWOR, GMG Ventures, Mercuri.vc, Moxxie Ventures.






