NocoBase

Quick Verdict

NocoBase is an AI plus no-code platform for business applications such as CRM, project management, ticketing, asset management, HR, and approvals. Operated by Singapore-based NocoBase Pte. Ltd., it does not ask AI to generate an entire opaque application from scratch. It provides relational data models, pages, roles, permissions, workflows, audit facilities, and a plugin framework, then lets people, external coding agents, and built-in AI employees work on that shared infrastructure. The project is actively maintained; GitHub listed v2.1.29 as the latest release on July 21, 2026.

It is a strong candidate for teams that want self-hosting, database control, and an extensible internal-system platform. One old claim needs correction: the current Community Edition is not an MIT/AGPL dual-licensed product. The NocoBase License Agreement updated February 24, 2026 incorporates Apache-2.0 but adds requirements to retain branding and intellectual-property information and prohibits using the original or modified software to provide a public no-code, low-code, or AI platform SaaS/PaaS. NocoBase markets the product as open source, but these additional restrictions mean legal teams should read the complete agreement before redistribution, white-labeling, customer delivery, or platform operation rather than relying on an Apache badge alone.

Best For

• Small and midsize organizations building self-hosted CRM, ticketing, projects, assets, approvals, operations dashboards, or other relational business systems.
• Implementation teams that need data modeling, field-level permissions, workflows, and plugins rather than a single collaborative spreadsheet.
• Developers and administrators able to operate Node.js, a relational database, containers, backups, upgrades, and plugin compatibility.
• Organizations that want coding agents to accelerate setup and extension while business owners inspect data models, pages, and permissions in a visual interface.
• It is less suitable for users who want a fully managed, maintenance-free SaaS, or for companies planning to offer a public low-code platform based on Community Edition. The latter is explicitly license-sensitive.

Key Features

• Data-model-first architecture: Define collections, fields, and relationships, then configure pages, blocks, actions, and APIs over shared relational data. The user interface is decoupled from the data model.
• WYSIWYG no-code building: Switch between usage and configuration modes so non-developers can refine forms, lists, dashboards, and interactions.
• Microkernel plugin system: Core features, commercial capabilities, and custom extensions are composed as plugins. Developers can extend the platform with TypeScript, Node.js, React, and Koa.
• Permissions and workflows: Roles, field-level read and write control, triggers, and many workflow nodes are built in. Approval, audit, SSO, version control, and clustering depend on edition and plugin.
• AI-assisted building: CLI and Skills support agents such as Claude Code, Codex, Cursor, OpenCode, and TRAE for setup, configuration, migration, plugin development, and release work.
• AI employees: Built-in workers can assist with analysis, Q&A, and form filling in the front end, or join back-end workflows for document recognition, risk monitoring, and task routing. They have roles and field permissions, and actions can be audited.
• Self-hosting and data sources: Community Edition can run in your environment with MySQL, PostgreSQL, or MariaDB as the main source. External databases and enterprise sources such as Oracle or ClickHouse vary by paid edition.

Use Cases

A typical project starts by modeling customers, orders, tickets, employees, or assets, then adding forms, lists, role permissions, and automated processes. Teams can implement specialized integration or logic through plugins. A coding agent can draft configuration, migration, and plugin code, but changes should remain on a branch, dependencies should be pinned, and people must review source and exercise migration and rollback paths. A green generated test log does not prove that relationships, authorization, transactions, or business rules are correct.

AI employees require even stricter operational controls because they can interact with live business data and workflows. Give every employee its own least-privilege role and restrict it to necessary collections and fields. Require human approval for writes, deletes, outbound messages, high-value transactions, and consequential workflow transitions. Store model API keys in the variables and secrets facility, never prompts or source. Document model providers, transmitted fields, logs, vector stores, retention and deletion, and verify that audit records identify each data change and workflow trigger.

Pricing

The official 2026 comparison lists Community Edition as free, Standard at a one-time $800, Professional at a one-time $8,000, and Enterprise by quote. Commercial licenses are perpetual and normally include a one-year upgrade and support period; pricing is not per seat, application, or data volume. Community includes unlimited users, applications, and records and most basic AI employee capabilities. Standard adds features such as branding replacement and selected external MySQL or PostgreSQL sources. Professional adds client-project rights, advanced AI employee capabilities, version control, SSO, and advanced workflows. Enterprise targets clustering, multiple applications and spaces, audit, observability, and additional enterprise data sources.

The matrix is detailed and changes with plugins, so verify the live page and contract. Community Edition can be used commercially within its agreement, but white-labeling, selling configurable applications to clients, or offering a platform service may require a different license or may remain prohibited. “One-time” also does not eliminate infrastructure, model API, backup, maintenance, and upgrade costs.

Pros

• Provides maintainable data, page, permission, and workflow infrastructure instead of one-off generated code.
• Community Edition is self-hosted and gives the organization control of its database without official per-user or per-record fees.
• Deep plugin architecture supports both visual configuration and TypeScript development.
• AI employees, external agents, MCP, HTTP APIs, CLI, and Skills create a broad AI extension path.
• Perpetual commercial licenses make the vendor cost model easier to forecast than a per-seat subscription.
• Active releases and a substantial public repository provide better operational visibility than a closed builder.

Cons

• Its additional license restrictions mean it should not be treated as unrestricted Apache-2.0 software; SaaS/PaaS and redistribution plans need legal review.
• Self-hosting transfers responsibility for databases, TLS, backups, recovery, vulnerabilities, upgrades, and plugin compatibility to the operator.
• Advanced SSO, audit, cluster, version-control, data-source, and AI employee features can require expensive commercial editions.
• An over-permissioned AI employee can amplify mistakes across records and workflows.
• Large plugin portfolios increase dependency, upgrade, regression-testing, and support costs.
• AI model traffic may still leave the self-hosted environment unless model and integration routes are deliberately controlled.

Alternatives

FAQ

What is the latest NocoBase release?

GitHub marked v2.1.29 as the latest release on July 21, 2026. Releases are frequent, so check the current release notes and plugin compatibility immediately before installation or upgrade.

Is NocoBase MIT or plain Apache-2.0 software?

It is not MIT. The current agreement incorporates Apache-2.0 and then adds NocoBase-specific branding, intellectual-property, and public platform SaaS/PaaS restrictions. The supplementary terms prevail where they conflict.

Can Community Edition be used commercially?

Yes, it may be used commercially and for internal applications with unlimited users, applications, and records, provided the additional terms are followed. White-labeling, client delivery, further configuration rights, and platform businesses require a careful license check.

Can an AI employee bypass user permissions?

The official design gives each AI employee a role and field-level permissions, with auditability. Administrators must still configure least privilege, test each role, and require human approval for writes, deletes, external actions, and other high-impact work.

Can NocoBase run on an enterprise intranet?

Yes, Community Edition is self-hostable, and paid editions add more advanced deployment capabilities. An internal deployment does not automatically keep AI data internal; inspect every connected LLM, MCP server, messaging integration, and API.

Do AI-built applications still require testing?

Yes. Independently verify relationships, every role’s read and write boundaries, migrations and rollbacks, retries, backups and restore, and plugin upgrades. Generated source and test output are evidence to inspect, not proof of correctness.

Bottom Line

NocoBase is compelling because it places AI-assisted building on durable relational data, pages, permissions, workflows, and plugins while allowing an organization to self-host and control its database. The v2.1.29 release confirms active development, and its current AI employee and coding-agent direction is much broader than its older no-code positioning. Adoption should begin with three checks: review the complete NocoBase license for the intended business model, price the required Community and commercial plugins, and govern AI employees as privileged system identities. With least privilege, secret management, audit, independent tests, backup and rollback, and human approval, it can support long-lived internal systems. For a temporary form or a team unwilling to operate infrastructure, a managed alternative is usually simpler.