Building the infrastructure layer for AI automation, from scratch.

Most teams building AI automation end up with the same mess, a tangle of scripts, hard-coded API calls, and no visibility into what's actually running. Every new workflow means re-implementing memory, retries, and error handling from scratch.

Linea is the platform we built to solve that properly.

It's a visual canvas where you compose workflows by connecting nodes, backed by a reliable execution engine and a TypeScript SDK so any application can trigger and stream results programmatically.

The node system covers 24 types across six categories, AI agents, data transforms, memory read/write, integrations (Slack, GitHub, Notion, Gmail), control flow, and infrastructure triggers. The Approval node pauses a workflow mid-execution and waits for a human to review before continuing. That's a first-class primitive, not a workaround.

Memory is persistent across executions. The Retriever node uses hybrid vector and BM25 search with RRF fusion, ranking by both semantic meaning and exact keywords simultaneously. AI is multi-provider: Anthropic, OpenAI, Groq, Google, and Ollama for fully local deployments.

The TypeScript SDK lets any codebase trigger workflows and stream execution events in real time. RBAC and audit logs are built in.

Next.js 15, React Flow, NestJS 11, LangGraph.js, BullMQ, PostgreSQL 16 + pgvector, Redis, Clerk, pnpm + Turborepo.

A complete, production platform., visual builder, execution engine, SDK, documentation, and user management, built from scratch and live at docs.getlinea.app.

Workflows that previously required weeks of custom engineering can now be composed visually and triggered from anywhere. Human oversight is built into the process rather than added as an afterthought.