What tech stack does Airtable use?

Airtable's backend runs on Node.js with TypeScript, migrated from a primarily JavaScript codebase — the Engineering Blog documented the migration of over one million lines of code to TypeScript (completed ~2022), followed by a second-stage optimization scaling TypeScript to thousands of isolated projects, with ~85% of projects now using isolated declarations for faster type-checking times. Production infrastructure runs on AWS, principally EC2, with Terraform for infrastructure-as-code and a Kubernetes-based compute platform that orchestrates all services — including AI inference workloads, vector databases, and document extraction pipelines introduced with the Superagent product.

For data infrastructure, Airtable uses Apache Kafka for event streaming, Apache Spark and Apache Flink for large-scale batch and streaming data processing, and Python for data science and machine learning workloads. This is a modern, cloud-native data stack consistent with a company processing billions of record operations daily across 500,000+ organizations. The AI infrastructure layer — multi-model LLM orchestration, agent coordination, and web-search tooling for Superagent — was added atop this foundation following the October 2025 DeepSky acquisition, using models from OpenAI, Anthropic, and Google simultaneously.

The frontend is a React/TypeScript single-page application, the canonical choice for Airtable's highly interactive, real-time collaborative grid interface. Mobile apps are native iOS and Android with JavaScript webviews where appropriate, consistent with a product that started web-first. The Engineering Blog's 2025 post on scaling TypeScript noted that many React components had complex inferred return types; Airtable wrote custom ESLint rules to automatically annotate them as React.ReactNode, improving compile performance significantly.

On analytics and experimentation, Airtable runs a multi-tool stack: Segment for event collection and routing, Mixpanel and Amplitude for product analytics, Google Analytics for web measurement, and Optimizely for A/B testing. FullStory provides session replay and digital experience analytics. Rollbar handles error monitoring. For GTM tooling, Intercom handles in-product customer communications and support, Marketo drives marketing automation, Mailgun handles transactional email delivery, and Mailchimp manages email marketing campaigns.

Airtable's all-in on AWS and Kubernetes signals strong receptivity to AWS Marketplace deals, cloud-native infrastructure tools, and Kubernetes-compatible observability or security products. Any tool that integrates with Segment (their event backbone) or surfaces data in Mixpanel/Amplitude has a natural integration story for Airtable's product analytics team. The TypeScript-first, isolated-declarations approach also signals receptivity to developer tooling that improves type-checking performance or monorepo tooling.

The AI infrastructure buildout — LLM inference, multi-model orchestration, agent evals, vector storage — is actively expanding with the Superagent launch and DeepSky acquisition. This opens budget for AI model providers, AI observability platforms (LLM tracing, evals), and enterprise AI governance tools. Airtable's use of OpenAI, Anthropic, and Google simultaneously makes them a candidate for multi-model routing or cost-optimization tools. On the GTM side, Intercom augmentation or alternatives is a real opportunity given the company's scale, as is any tool that adds intent data or enrichment signals to the existing Marketo-based RevOps stack.