MeDo

What MeDo actually is

MeDo is an AI app builder with an unusually complete definition of "app". You describe a product in one plain-language sentence — the site's own example is "build a booking app for my fitness studio with memberships and class schedules" — and the platform generates a working web or mobile application with an interface, authentication, backend logic, a connected Postgres database, and a permanent live URL on its medo.site domain. The stack underneath is conventional and readable: React (Vite + TypeScript) on the frontend, Supabase on the backend (Postgres with row-level security, Edge Functions, Storage), OAuth plus email sign-in, Stripe-ready payment wiring, webhooks, custom domains with HTTPS, and logs and error monitoring. This is not a wireframe generator; the pitch is that one prompt yields one deployed, editable product.

The product launched on Product Hunt in November 2025 under the banner "MeDo by Baidu", where it took #1 Product of the Day with 514 upvotes and 146 comments (hunted by Chris Messina), and a second launch followed in February 2026 under the tagline "agentic AI builds full-stack apps in minutes". Identity is where the record gets thin: Product Hunt associates the product with Baidu, while LinkedIn shows a company with around two employees and no published headquarters. We could not establish who legally operates medo.dev, where it is based, or how it relates to Baidu beyond that branding — a caveat we will return to.

What is documented is the product surface. MeDo claims generation across 29 formats and 8 platforms — websites, multi-page web apps and SaaS-style products with auth and dashboards, cross-platform mobile apps, iOS and Android builds described as store-ready, H5/PWA, desktop, and browser extensions — plus smaller artifacts (slide decks, chatbots, forms, wireframes). Every build comes with the same plumbing: accounts and permissions generated with the product, backend APIs created automatically, a connected database, and a permanent free address that stays editable after publishing. The homepage template gallery shows real live examples — studio booking apps, fitness member portals, agency operations tools — which is more convincing evidence than any feature grid.

Pricing: what you really pay

Public pricing stops at two visible tiers before the site asks you to sign in. Free is $0: 400 credits per month delivered as login bonuses, one video and five image generations per day, a permanent free MeDo URL, and 50 MB single-file uploads per project. Pro is $20 per month. A "Max" tier exists in the documentation — code download is described as "a Pro and Max feature" — but its price is behind the login wall, as is the full credit-system breakdown. We dislike pricing pages that require an account before showing their own price list, and we note it here.

The credit system is the real meter. Everything costs credits, including the one-off items: downloading your app's source is 300 credits per download. On the Free tier, where credits arrive as monthly login bonuses, a single code download is therefore a meaningful fraction of your monthly allowance — the design pushes you toward Pro long before you have outgrown the product functionally. The docs are at least explicit about costs ("budget an hour for a first migration"), which we count as honesty even when the metering is aggressive.

The escape hatch is better than most AI builders offer. The export documentation walks through downloading the full source as a ZIP — frontend code, Supabase schema with RLS policies and triggers, migration scripts, and Edge Functions — and migrating it to your own Supabase project, with a checklist, tool requirements (Node 20+, Supabase CLI, psql), and blunt warnings about what is not included (row data, stored files, secrets). If you plan to self-host eventually, budget for a VPS or platform to run it on — our /category/vps-providers listings are a starting point.

Performance & infrastructure

MeDo is a SaaS platform, so there is no hardware sheet to grade, and the company publishes no datacenter map, no regions, and no uptime figures. The underlying runtime is Supabase-based, which means your app's data lives on Supabase infrastructure wherever MeDo hosts it, and the export path confirms this: the shipped .env file points at MeDo's hosted Supabase URL and anon key until you swap them. For anyone with data-residency requirements, that unanswered question — which region, which jurisdiction, which legal entity — is the first thing to ask before storing anything real.

On generated-app quality, the documentation deserves credit for candour. It states exactly what a download contains (structure and function code, no data, no keys), that MeDo replaces the npm dev and build scripts with placeholders so you must run Vite directly, and that most users cannot export row data from the source database at all. That level of migration documentation is rare in this category and suggests engineering culture rather than pure marketing.

We have not benchmarked generated apps, hosting performance, or build times — no hands-on testing claims here, editorial judgment only. The platform's own statistics (350K+ active users, 550K+ apps created, 6M+ app views) are vendor-published and unverifiable; the one externally checkable claim, the Product Hunt award, checks out. Treat the rest as marketing until audited.

Where it falls short

Corporate opacity is the big one. A product that stores your customers' data should tell you who owns it. Between "MeDo by Baidu" on Product Hunt and a two-person LinkedIn listing, we cannot tell whether this is a Baidu product, a startup licensing a name, or something in between — and the Terms of Service and Privacy Policy live in the docs rather than answering the ownership question. For hobby projects this is tolerable; for client work it is a procurement problem.

The platform answers its own questions inconsistently. In one sample transcript, the in-app assistant tells a user "I don't generate downloadable source code or zip files", while the documentation describes a paid Download Code feature in detail. A user's first impression of what is possible depends on which surface they asked.

Lock-in is designed around Supabase. The export path is real but narrow: Vite/React frontend, Supabase backend, your own migration effort for data. If you wanted a different backend, or your app outgrows the credit model, the answer is a rewrite of the wiring rather than a config change. And credit metering with 300-credit downloads, daily generation caps, and monthly login bonuses is fiddly enough that cost forecasting requires reading the docs closely.

Finally, the reliability record is simply absent: launched in late 2025, no published SLA, no status history we could find, no independent long-term reviews. Early adopters should version-control their exports and keep an exit path warm.

Who should pick this

Our pick for founders and small teams prototyping internal tools, client booking flows, and MVPs who want a deployed product rather than a mockup, and who are comfortable on a young platform. The single-prompt-to-live-URL flow is genuinely fast, and the fact that the generated product includes auth, a database, and a real deployment removes the weekend of plumbing that usually follows a prototype.

Students, educators, and hobbyists get real value from the free tier — the permanent URL and 400 monthly credits are enough to learn on and publish simple tools.

We would not pick MeDo for: anything with regulated or residency-bound data (ownership and hosting region are unresolved), production systems needing an SLA, teams that need a portable multi-cloud backend, or agencies whose clients require vendor due diligence. For those cases, build on infrastructure you can name — and when you outgrow hosted builders, running the exported stack yourself on a solid VPS is the honest fallback.

Key takeaways

• MeDo turns one plain-language prompt into a deployed web or mobile app with React frontend, Supabase backend (Postgres + RLS), auth, payments wiring, and a permanent free URL.
• Launched on Product Hunt in November 2025 as "MeDo by Baidu" and took #1 Product of the Day (514 upvotes); corporate ownership and HQ remain unverified.
• Pricing: Free ($0, 400 monthly credits, 1 video + 5 image generations/day) and Pro ($20/month); a Max tier exists but its price sits behind the login wall.
• Code export is real (300 credits per download, Pro/Max only) with detailed self-hosting docs — but no row data, files, or secrets are included, and the migration target is Supabase.
• Documentation is unusually candid about limits; the in-app assistant, by contrast, has denied that code export exists at all.
• Good for MVPs and internal tools; not yet a defensible choice for regulated data or SLA-dependent production.

Sources

• https://medo.dev/
• https://docs.medo.dev/10-download-and-building-the-app.html
• https://producthunt.com/products/medo-2
• https://hunted.space/dashboard/medo-2/launches/medo-by-baidu
• https://cards.keenable.ai/w/01a16874-5ed4-7bb3-a249-76515fd00be6