Zoer

Positioning

Zoer sits in the most crowded, most over-promised corner of AI tooling: the prompt-to-app builder. Our catalog frames it as "generate production web apps from database schemas", but that description gets the arrow pointing the wrong way. In my reading of Zoer's own docs and quick-start guide, the flow is the opposite: you describe an application in one plain sentence, and Zoer designs the database schema for you, writes the backend CRUD APIs, builds the React frontend, and deploys it. The schema is an output, not an input. Where databases enter from outside is as connection targets — the platform advertises native PostgreSQL plus connections to MySQL, MongoDB and Supabase — and as a managed default called "Zoer Database" that requires no setup. If you already own a schema and want an admin generated from it, that is a different product category (Retool-likes and schema-driven scaffolders), and Zoer is not it.

Once that is clear, the pitch is genuinely ambitious for the price: a non-technical founder types "Create a dashboard to manage skincare customers" and receives a deployed, database-backed application they can modify in chat, export as Next.js + TypeScript source, and publish to a community marketplace with an optional price tag. For solo builders and small businesses, the draw is obvious. The question this review tries to answer is how much of that survives contact with real use — and Zoer's own troubleshooting documentation turns out to be unusually revealing on that point.

What it actually does

The build loop, per the quick-start: describe your idea in one sentence, press "Enhance Prompt" to expand it into a detailed specification, configure project settings (public or private visibility, database choice), then "Build App". Public apps can be featured in the community gallery; private ones stay hidden. If you enable Zoer Database, the AI analyzes your requirements, generates tables, relationships and constraints as a SQL script (the blog shows a CRM example producing companies, contacts, deals and activities tables in PostgreSQL), and pre-populates sample rows so the app is testable immediately. After the build, changes are conversational — "add a contact form to the homepage", "change all buttons to blue" — and an AI Copilot side panel in the deployed app handles data management in natural language ("create a new user record for John Smith…"), with a review step before operations execute.

Export matters here and Zoer offers it: you can download the complete source (the blog describes a Next.js application with TypeScript, full API routes, and an app.sql PostgreSQL schema) and continue outside the platform. That is the single most important feature in this category, because it converts a rented prototype into something you own. The commerce side — publishing to the community and monetizing by setting a price — exists in the docs, though I found no evidence of a mature marketplace yet.

Pricing, from zoer.ai/pricing at time of writing, is credit-based. The page I read shows three paid tiers: Starter at $8/month as displayed (a labelled special offer, 300 credits per month, up to 30 apps across public and private, free Zoer database, branding removal, code export, web publishing, standard email support); Pro at $30/month monthly or $24/month on the yearly toggle (20% off, 1,000 credits, unlimited apps, unlimited database records, custom domain, priority support); Max at $99/month monthly or $80/month yearly (3,500 credits, prioritized access to new features, free domain certificate, dedicated support). Credits are consumed when building and modifying applications; complex modifications cost more, and the FAQ says failed builds typically don't consume credits. Directory listings (toolify, creati, toolstory, software informer) still describe an older structure — a free plan with 15 credits and three public apps, Starter at $19 with 600 credits — so the plan shapes have shifted at least once and the current page did not show me a free card. Check the pricing page before you budget.

My testing notes

The most useful thing I did here was read Zoer's own FAQ end to end, because it documents the product's failure modes in more detail than most vendors admit. Take the preview-versus-build distinction: preview mode has high error tolerance and will render an app with minor defects, while build mode runs syntax validation, type checking and dependency resolution and fails on anything production-hostile. The FAQ then lists the recurring culprits in AI-generated code — wrong variable types, undeclared libraries, lint rules treated as build errors, Node.js APIs used in browser code, broken imports. In my experience that list is exactly what separates a demo from a deployable app, and the fact that users hit these often enough to need a documented paste-the-error-back-to-the-AS workflow tells you the build loop is iterative rather than one-shot. The prescribed fix — copy the full error, paste it into chat with the code, ask for a corrected version — works, but it means your first build is a starting point, not a finish line.

Second observation, and my favourite concrete detail in the whole docs set: the AI regularly writes an X-Frame-Options: DENY security header into the generated middleware, which then breaks Zoer's own iframe-based preview pane — the preview shows "Refused to Connect". The official remedy is to tell the AI, verbatim, "do not deny iframe connections". The generator is smart enough to add security headers and not smart enough to know its host platform needs iframes; the user sits between the two. I have seen this class of self-inflicted conflict in several prompt-to-app tools, and it is a good litmus test for how much of the stack the vendor actually controls.

Third, the attachment handling is brittle in a way that costs sessions. Upload a ZIP, a PDF, or any non-image file as chat context and requirement analysis fails — and because the bad file persists in the conversation history, every subsequent message in that chat fails too. The only documented fix is abandoning the chat and starting a new one. Only PNG, JPG, JPEG and SVG are accepted, and only as design references. That is a small implementation detail with an outsized annoyance factor, and it is exactly the kind of edge a first-time user hits on day one.

What it does not do

User authentication does not exist yet — the FAQ states registration, login, password and session handling are "under development", despite a Zoer blog post advertising "built-in Google & email auth" for generated CRMs. Trust the FAQ; the blog is selling the destination. Custom domains are contradictory the same way: the Pro tier lists "Custom Domain" as a paid feature while the FAQ says custom domain deployment is "under development" — I could not confirm which is current. There is no visible free tier on the pricing page as I read it. The credit system is opaque — the site never publishes what a given build or modification costs in credits, so budgeting is guesswork. Nothing in the product replaces engineering judgment: apps that preview fine can fail strict builds, security posture is whatever the model happens to write, and testing beyond the pre-populated sample rows is on you. And while code export exists, the exported app's production readiness is unverified — I found no independent review validating the exported code quality.

Verdict

Zoer is a real full-stack generator at a starter price that undercuts most of its category, with the two features I weight most heavily — schema generation with sample data, and complete code export — both present and documented. It is also visibly young: authentication is missing, custom domains are contradicted by its own docs, credit costs are unpublished, and its FAQ reads like a war diary of AI-code failure modes. For a founder prototyping an internal tool or a small-business dashboard, the Starter tier is cheap enough to experiment with and the export feature means you are not trapped. For anything touching real users and real data, plan for build-fix iteration and your own security review. I rate Zoer 3.2 — the depth of what it generates and the ownership export carry the core-functionality half of my rubric (40%), the price-to-output ratio is strong (20%), usability is held back by the preview/build friction and attachment traps (20%), support via Discord and email is present but small-team (10%), and reliability signals are thin with marketing figures ("15,000+ systems built", "98% deployment success rate") that remain unverifiable vendor claims (10%).

Sources

• https://zoer.ai/docs/start-guide/getting-started
• https://zoer.ai/docs/start-guide/faq
• https://zoer.ai/docs/core-features/zoer-database
• https://zoer.ai/pricing
• https://zoer.ai/blog/ai-crm-builder-guide
• https://zoer.ai/pages/solutions/website/small-business-website.html
• https://zoer.ai/posts/zoer/ai-shopify-website-builder-ecommerce-guide
• https://toolify.ai/tool/zoer
• https://creati.ai/ai-tools/zoer
• https://toolstory.ai/tools/70131/zoer
• https://zoer.software.informer.com