APITable

APITable is an open-source, API-oriented low-code platform built for teams who need a collaborative database with the flexibility of a spreadsheet and the power of a full REST API. It solves the problem of fragmented tooling by unifying data storage, real-time collaboration, automation, and external integrations into a single self-hostable product — making it a direct Airtable alternative you control.

Every table in APITable automatically exposes a complete Data and Metadata API, removing the need to hand-build backend endpoints for common internal-tool workflows. The platform supports 7 view types (Grid, Gallery, Kanban, Mindmap, Gantt, Calendar, and Form), unlimited cross-table links, an embeddable widget system, built-in robot automation, and integrations with Zapier, Make, n8n, and Appsmith.

The technology foundation is a monorepo spanning TypeScript (Next.js for the frontend, NestJS for the real-time Room Server) and Java (Spring Boot for the backend API server), with MySQL for relational storage, Redis for caching and presence, and MinIO for object storage. All services are orchestrated via Docker Compose, enabling one-command self-hosted deployment.

Enterprise-grade features — SAML SSO, audit logs, row/column-level permissions via Mirror views, database auto-backup, watermarks, and a Datasheet Query Language (DQL) for SQL-like queries — are all included in the open-source build, giving self-hosters the same feature surface as the cloud offering.

What You Get

• Realtime Collaboration with OT Algorithm - Multiple users can edit the same table simultaneously using Operational Transformation for conflict-free updates, maintaining consistency across 100k+ row datasets with smooth canvas-rendered performance.
• Automatic API Generation - Every datasheet automatically exposes a full REST API for both Data and Metadata, complete with a one-click in-app API panel that generates ready-to-run code snippets for common languages.
• Unlimited Cross-Table Links - Create bidirectional and unidirectional relationships between any tables across workspaces without limits, enabling complex data models like CRM pipelines, ERP systems, and project hierarchies.
• 7 View Types - Grid, Gallery, Kanban, Mindmap, Gantt, Calendar, and Form views are each fully customizable with independent filters, sorting, grouping, and field visibility settings per view.
• Embeddable Tables and Dashboards - Generate embed HTML scripts to insert live interactive datasheets directly into websites, help centers, or internal portals with no additional backend required.
• Built-in Robot Automation - Configure trigger-action automations — Slack messages, emails, webhook calls, or custom robot actions — fired on form submissions, row changes, or scheduled intervals.
• Row and Column Permissions via Mirror Views - Implement row-level access control by creating Mirror views that filter data by team or user context, combined with per-column permission toggles.
• SSO and SAML Integration - Enterprise authentication via SAML 2.0 and Single Sign-On is included in the open-source edition, enabling corporate identity provider integration without a separate license.
• Datasheet Query Language (DQL) - Query tables using a SQL-like syntax with joins, filters, and aggregations — accessible from both the UI and the REST API for advanced data retrieval.
• Widget System with 20+ Open-Source Widgets - Extend dashboards with customizable chart, formula, and data visualization widgets, all available as open-source and customizable per project.
• Database Auto-Backup and Data Exporter - Schedule automatic backups and export data in multiple formats to prevent data loss in self-hosted production deployments.
• AI Copilot and Automation - Chat with your database to generate charts, build automation workflows, analyze data, or create chatbots using built-in AI capabilities available on the cloud edition.

Common Use Cases

• Building a no-code CRM - A sales team tracks leads, links contacts to deal pipelines, embeds intake forms on their website, and pushes new records automatically to their email tool via Zapier — all without a backend developer.
• Running project management across teams - A product team uses Kanban and Gantt views to manage sprints, assigns tasks with deadlines, links issues to a documentation datasheet, and shares live embed views in Confluence pages.
• Replacing a fragmented SaaS stack - An operations team consolidates vendor tracking, purchase orders, and inventory into linked APITable datasheets, then queries across them with DQL and exposes the data through the REST API to a custom reporting dashboard.
• Embedding live data into a customer portal - A SaaS company embeds customer-specific APITable views into their help center so users can see their subscription status, usage data, and support tickets in real time, with row permissions filtering each customer’s data.
• Automating internal approval workflows - An HR team uses form views to collect time-off requests, triggers a robot automation to notify managers in Slack, and tracks approval status in a linked table with audit logs.

Under The Hood

Architecture APITable is organized as a monorepo containing distinct service packages that map to separate runtime components: a Java Spring Boot backend server handling business logic and persistence, a NestJS room server managing real-time collaboration and WebSocket state, a Next.js web frontend, a databus server for high-throughput data transport, and an API gateway layer routing between them. The real-time layer uses a dedicated Operational Transformation engine in the core TypeScript package to reconcile concurrent edits against a Changeset/Operation/Snapshot model, keeping collaboration correct under network partition and multi-user conflict scenarios. The separation between the collaboration engine and the persistence layer means write fanout and read queries are handled by different services, avoiding the coordination bottlenecks common in monolithic collaborative tools. All services are composed via Docker Compose with explicit health checks and startup ordering, reflecting a microservices approach that trades operational simplicity for deployment flexibility.

Tech Stack The frontend is built with React 18 and TypeScript inside a Next.js application, rendered via a canvas-based grid engine for high-performance datasheet display rather than DOM-heavy table rendering. The real-time server is NestJS (TypeScript) using TypeORM against MySQL, with Redis handling presence, pub/sub, and API response caching. The backend REST API is a Spring Boot Java service using MyBatis for database access, exposing data and metadata endpoints consumed by both the frontend and external API clients. MinIO provides S3-compatible object storage for attachments and exports. The build system is an Nx monorepo with pnpm workspaces, and CI is handled via GitHub Actions. Cypress covers end-to-end tests while Jest handles unit and integration coverage for both TypeScript packages.

Code Quality Test coverage is present and meaningful across the codebase — the TypeScript core package has dedicated test suites for the formula parser, automation trigger logic, and OT engine edge cases, while the NestJS room server has service-level spec files with realistic test fixtures. The Java backend has integration test infrastructure via Spring Boot test context with abstract base classes for database-backed tests. TypeScript strict mode is enforced across packages with consistent interface segregation between domain models and API response types. Error handling in the NestJS layer follows typed exception patterns with NestJS’s exception filters, though the Java layer uses a mix of custom exception hierarchies and Spring’s standard error propagation. Linting and formatting are enforced via ESLint and Prettier with monorepo-wide configuration, and commit message conventions (Conventional Commits) are enforced via commitlint.

What Makes It Unique The most technically distinctive aspect of APITable is its canvas-based grid rendering engine, which bypasses DOM virtualization entirely to achieve smooth scrolling and editing at 100k+ rows — a deliberate architectural choice that separates it from DOM-table-based Airtable clones. The OT collaboration engine is implemented in pure TypeScript in the shared core package and shared between the frontend and the room server, meaning conflict resolution logic runs in the same implementation on both sides of the wire. The Datasheet Query Language (DQL) exposes a SQL-like interface directly over the datasheet model, enabling cross-table joins that the underlying schemaless spreadsheet model doesn’t natively express. The Mirror view system implementing row-level permissions through filtered views rather than a separate ACL layer is an unusual design choice that makes permission-aware data sharing composable without schema changes.

Self-Hosting

APITable is released under the GNU Affero General Public License v3.0 (AGPL-3.0), one of the strongest copyleft licenses available. For self-hosters, AGPL means you can run APITable freely for internal use — including commercially — without open-sourcing your own code. The restriction applies if you distribute a modified version of APITable as a network-accessible service: in that case, you must make your modifications available under AGPL. There is one practical edge case called out in the project’s LICENSING.md: if you embed APITable in an external-facing application and remove the “Powered by APITable” branding, the embedding application itself would fall under AGPL unless you purchase a commercial Advanced Embedding License from the team.

Running APITable yourself is non-trivial from an infrastructure standpoint. The Docker Compose stack includes roughly a dozen services: a web server, backend API server, room server (NestJS), databus server, API gateway (Nginx), MySQL, Redis, MinIO, RabbitMQ, and initialization jobs. The recommended minimum is 4 CPUs and 8 GB RAM for a functional single-node deployment. You are responsible for MySQL backups, Redis persistence configuration, MinIO storage durability, and coordinating upgrades across all container images — there is no managed upgrade path in the open-source edition. The all-in-one Docker image is available for demos but explicitly not recommended for production or enterprise use.

Compared to the cloud-hosted aitable.ai offering, self-hosters get the same core feature set including SAML SSO, audit logs, and the automation robot system. What the managed cloud adds is operational reliability — managed uptime, automatic database backups handled by the provider, rolling upgrades without service interruption, and direct enterprise support with SLAs. The community Discord and GitHub Issues are the primary support channels for self-hosted deployments. Teams evaluating self-hosting should factor in the operational overhead of a multi-service stateful system before committing, particularly if they lack dedicated infrastructure engineering capacity.