n8n: free self-hosted, 70+ AI nodes, LangChain, n8n-mcp (15.4K stars). Zapier: 8,000+ apps, no-code, task-based pricing. n8n wins for developers/AI agent builders. Zapier wins for non-technical business teams needing maximum breadth.
Automation
Three sub-categories with distinct buyers: Workflow Automation (n8n #1, Activepieces, Zapier), Code-First Orchestration (Windmill, Trigger.dev, Inngest, Kestra), Agent Integration (Composio, Pipedream MCP). n8n dominates overall with 180K stars, n8n-mcp (15.4K stars), and dedicated Claude Code skills. Inngest leads npm downloads (499K/wk). Composio is provisional — zero HN traction despite 27K stars.
10
Ranked
8
Signals
Current ranking
Best for: Technical teams building complex, multi-step automations with AI agents — the default choice for Claude Code users needing workflow automation
180K stars, 571 commits/30d, 583 contributors. n8n-mcp (15.4K stars, 265 AI-capable tool variants). n8n-skills (7 Claude Code skills). Consensus #1 across 9+ independent comparison sites. $40M ARR, 3,000+ enterprise customers. 70+ AI-native nodes with LangChain integration.

Best for: AI agent developers needing managed auth across dozens of services — use inside n8n or alongside other workflow tools
27K stars, 857 commits/30d (highest velocity). 850+ integrations via hosted MCP servers with managed OAuth. Architecturally differentiated as integration layer FOR agents, not a workflow builder.

Best for: Data engineering and DevOps teams migrating from Airflow — YAML-declarative workflows with Git version control
26.5K stars, 433 contributors, v1.3.3 (2026-03-17). Airflow 2 EOL creates structural migration demand. 600+ plugins, multi-language support.

Best for: Startups and non-developer teams wanting open-source workflow automation with simplicity
21K stars, 419 contributors. Unique: $25/mo flat pricing with unlimited task executions. YC S22. Native AI steps (Claude, OpenAI). Independent comparisons position as 'easier n8n.'

Best for: TypeScript developers needing background jobs with long-running compute or AI agent workflows
745pt HN peak (highest in category). 258K npm downloads/wk (real adoption). Apache 2.0 license. Transparent security incident post-mortem (262pts) builds trust.

Best for: Next.js/Vercel teams needing serverless workflow orchestration with AI inference built in
499K npm downloads/wk — highest in automation category by 2x. Independent confirmation as 'better default for typical Next.js SaaS' (HashBuilds). step.ai.infer() API offloads LLM requests.

Best for: Developers wanting workflows-as-code — internal tooling, DevOps automation, script orchestration
428pt + 378pt HN peaks — strong developer reception. Code-first (Python/TypeScript/Go/Bash) with auto-generated UIs. 420 commits/30d, v1.661.0 extreme release velocity. Independent comparisons confirm best for code-heavy internal tooling.

Best for: Quick MCP access to a massive API surface without building your own auth — one-off integrations
3,000+ APIs via MCP with 10,000+ prebuilt tools. 10K free invocations/month. Credentials encrypted at rest. Featured in 'Best MCP Servers for Automation 2026' (Fast.io).

Best for: Non-technical business teams needing the broadest integration surface — 8,000+ apps, 40,000+ actions
8,000+ integrations — unmatched breadth. Official MCP server. AI Copilot for natural language automation. For business users who need breadth without touching code, Zapier is the answer.

Best for: Track only — re-evaluate Q3 2026 when developer adoption signals emerge
$50M Series B from Benchmark (March 2026), $70M total. YC W24. Customers: Shopify, Ramp, Instacart. AI-native architecture is genuinely interesting.

Head to head
n8n: 8x community (180K vs 21K), 5x integrations (1,000+ vs 200+), MCP bridge, Claude Code skills. Activepieces: easier onboarding, $25/mo unlimited tasks. n8n for power users; Activepieces for non-technical teams wanting open-source.
Inngest: 499K npm/wk (2x Trigger.dev), serverless-native, Next.js default. Trigger.dev: Apache 2.0 (self-hostable), 745pt HN, heavy compute. Inngest for serverless/Vercel; Trigger.dev for long-running compute exceeding serverless limits.
Different layers — Composio is an agent auth/integration layer, not a workflow builder. Composio solves managed OAuth for 850+ apps; n8n/Zapier build visual workflows. Complementary, not competing. Use Composio inside n8n workflows.
Kestra: YAML-declarative, Java, data engineering/DevOps, Airflow migration target. n8n: visual canvas, TypeScript, business process/AI. Kestra for data pipelines/ETL; n8n for business automation with AI nodes. Different buyers.
Windmill: code-first (Python/TS/Go/Bash), auto-generated UIs, internal tooling. n8n: visual canvas, 1,000+ integrations, AI nodes. Windmill for 'run scripts in sequence with a UI'; n8n for integration-heavy visual workflows.
Pipedream: 3,000+ apps via MCP, 10K free invocations, developer-centric, encrypted credentials. Zapier: 8,000+ apps, no-install OAuth. Pipedream for developer MCP access; Zapier for non-technical business users.
Public signals
n8n-mcp provides Claude Code with access to 1,239 n8n workflow nodes — 99% node property coverage, 265 AI-capable tool variants. The highest-starred MCP automation tool. Same author ships n8n-skills (7 Claude Code skills).
Cross-validated across OpenAlternative, Hatchworks, Digidop, HostAdvice, and 5+ more independent comparison sites. $40M ARR, 3,000+ enterprise customers (Cisco, Microsoft, Liberty Mutual). 728pt HN peak sustained over 6+ years.
27,441 stars and 857 commits/30d but zero HN posts above 10 points. Star-to-engagement ratio is the most anomalous in the category. Architecture is genuinely differentiated (agent auth layer), but evidence base is thinner than star count suggests.
Every Airflow 2 user needs a migration target. Kestra's YAML-declarative approach with 600+ plugins is the closest match. 26.5K stars, 433 contributors.
Most-installed automation tool by 2x margin. Downloads dwarf star count, suggesting deep practical adoption in the Vercel/Next.js ecosystem. Independent: 'for typical Next.js SaaS, Inngest is the better default' (HashBuilds).
Highest HN peak in automation (745pts, 190 comments). Transparent security incident post-mortem (262pts) builds community trust. 258K npm downloads/wk confirms real adoption.
Despite 13.8K stars and strong HN launch (317pts), zero commits in 30 days and last release August 2025. Project appears abandoned. Do not recommend.
Benchmark-led $50M at this stage signals strong investor conviction. Customers: Shopify, Ramp, Instacart. But zero public GitHub, zero HN, zero developer community. Re-evaluate when developer adoption signals emerge.
What changes this
If Composio gets independent HN/community validation, moves from provisional #2 to solid #2. If star farming confirmed, drops to #5-6 and flagged as untrusted.
If n8n ships an official (first-party) MCP server, cements #1 further — n8n-mcp community project becomes less critical.
If Gumloop ships public API / developer tools, enters ranking at #4-6 depending on quality.
If Inngest adds self-hosting, moves up 1-2 ranks.
If Zapier drops task-based pricing, moves up significantly for developer audience.
If Kestra ships an MCP bridge, moves up 1 rank.
If Automatisch resumes development, re-enters ranking at #7-9.
If Pipedream addresses issue backlog (4,362 open), moves up 1-2 ranks.
If Browser-Use ecosystem matures MCP bridge, new sub-category entry (Browser Use + Skyvern) joins ranking.