metatable.ai

What It Is

Metatable is a “spec-driven” AI software generation tool. It is positioned as neither traditional no-code nor a simple prompt-to-app builder. Users describe business processes in natural language; the system first generates a reviewable, structured PRD/specification, then uses that spec to generate the database, backend, frontend, APIs, deployment path, and exportable source code. The official site emphasizes that its target users are business teams who cannot safely assess code, so it treats “business-readable specifications” as the source of truth before code generation.

Core Capabilities and Outputs

Its generation stack is clearly defined: PostgreSQL database, Rust backend, Flutter frontend, along with typed APIs, authentication flows, deployment validation, and source code export. Compared with development environments such as Claude Code and Cursor, Metatable aims to reduce users’ dependence on codebases, terminals, Git, testing, and deployment. Compared with prompt-based app generators such as Lovable and Bolt, it focuses on using specifications to lock down roles, records, permissions, workflows, and edge cases, reducing the risk of logic drifting after multiple rounds of prompting.

On the technical side, the official site mentions context engineering, DAG-based dependency ordering, executable validation, and self-fixing loops: SQL is run against a real database, Rust is actually compiled, and the system then fixes itself based on database or compilation errors. This suggests a focus on runnable software rather than just generating UI demos. However, the site does not disclose the underlying large language model.

Pricing and Trial

The Free plan costs $0/month, requires no credit card to start, and includes 1 project, 250k AI tokens/month, 1GB storage, and 1k requests/month. It is suitable for validating a single app idea, but it does not include backups. Standard is $25/month and includes 5 projects, continuous hosting, project export, and 90-day backups. Pro is $79/month and supports up to 10 users, 25 projects, and 365-day backups. Enterprise starts at $700/month and includes dedicated instances, human-assisted development, and custom architecture discussions.

Pros, Cons, and Best-Fit Users

The main advantage is its clear methodology: specs first, code second. It is well suited to business software with data models, permissions, and processes, such as CRM systems, inventory tools, internal tools, approval workflows, operations dashboards, and customer/supplier portals. Source code export also helps reduce lock-in risk. The limitations are also fairly clear: frontend generation and iteration are still maturing, and it is not suitable for games, highly customized consumer-facing interfaces, or products where the frontend experience is the core value. The official site reports a 95% database generation success rate, but a 72% backend deployment success rate (October 2025 funnel, 13 of 18 builds), so complex projects may still require human intervention.

Access from China, Payments, and Alternatives

The captured text does not clarify availability from mainland China, payment methods, Chinese UI, or Chinese customer support; it only mentions a Plain-English requirements chat, so Chinese-language support should be considered unknown. For adoption by teams in China, it would be necessary to first verify access stability, the usability of overseas payment methods, and whether generated code can be migrated to local infrastructure or a private cloud. Alternatives include Claude Code, Cursor, Lovable, Bolt, and various no-code/internal tool builders. If specifications, source code ownership, and backend consistency are priorities, Metatable is more differentiated.