Open source · MIT · Running on Cloudflare

Visual workflow
automation on Cloudflare

Build workflows visually by connecting artificial intelligence (AI) models, browsers, data pipelines and APIs, then deploy them on Cloudflare's serverless platform. Workflows scale to zero when idle and up with demand, and the entire stack is yours under MIT. No enterprise tier, no fair-code carve-outs, no asterisks.

Automate something

By signing up, you agree to our Terms of Service and Privacy Policy.

Built on · Integrates with

CloudflareAnthropicOpenAIGoogleGitHubDiscord

01 — Overview

Serverless workflows, zero infrastructure

Scale to zero when idle, to whatever traffic throws at you when it's not

Visual Workflow Editor

A React Flow workflow builder for creating automation workflows that makes command-line enthusiasts mildly uncomfortable. Build workflow pipelines by connecting nodes visually. No infrastructure setup, no Docker containers. Just workflows that run on Workers.

React FlowDrag & dropLive previewVersioning

Durable Workflow Execution

Run workflows on Cloudflare Workflows and Workers, where servers are merely a philosophical concept. Executions scale to zero when idle and up to whatever traffic throws at you, with no infrastructure to provision. Durability is built in, so long workflows survive restarts and retries.

Cloudflare WorkersWorkflowsDurable ObjectsAuto-retry

Persistent Workflow Storage

Save workflow state, execution history, and automation data using D1 SQL databases, R2 object storage and Workers Analytics Engine with reasonable confidence they'll still be there tomorrow. Durable workflow execution that persists even when things go sideways. Because workflows should complete, eventually.

D1 SQLR2 StorageAnalytics EngineKV

Workflow Triggers & Queues

Nothing happens unless something happens. Trigger workflow automation via HTTP webhooks for event-driven workflows, Queues for reliable message processing, scheduled cron jobs for time-based automation, or manual triggers when you feel like it. Connect to any REST API, integrate with third-party services, and orchestrate complex automation pipelines.

WebhooksQueuesCronManualEmail

470+

Workflow nodes

30+

AI models

12+

Integrations

MIT

Open source license

02 — Use cases

Automation use cases

What you can build with workflow automation

Browse all workflow templates →
Image Generation
imagegenerationai

Generate images from text prompts using AI models like Flux, Stable Diffusion, and Imagen. Create unique visuals for marketing campaigns, social media posts, blog illustrations, or creative projects. Build automated pipelines that combine text generation with image creation, allowing you to produce complete visual content from simple inputs. Supports various styles, aspect ratios, and output formats for different use cases from thumbnails to high-resolution artwork.

Speech to Text
audiotranscriptionai

Transcribe audio files to text using Whisper AI with support for multiple languages and accents. Convert podcasts, meeting recordings, interviews, voice memos, and any audio content into searchable, editable text. Build automated workflows that process incoming audio files, generate accurate transcripts, and pipe the text into downstream nodes for summarization, translation, or analysis. Perfect for podcast production, meeting documentation, content repurposing, and accessibility compliance.

Text Summarization
textsummarizationai

Summarize long text content using AI models like Claude, GPT, and Gemini. Extract key points and condense articles, documents, reports, or any text into concise summaries. Build workflows that automatically process incoming content, identify the most important information, and generate summaries tailored to different audiences or formats. Perfect for content curation, research pipelines, news aggregation, and information processing workflows that need to handle large volumes of text efficiently.

Text Translation
texttranslationai

Translate text between over 100 languages using AI-powered translation models. Automate content localization for websites, apps, marketing materials, and customer communications. Build multilingual workflows that accept text in any language, detect the source language automatically, and produce accurate translations in one or more target languages. Ideal for international businesses, content localization pipelines, multilingual customer support, and cross-border communication workflows that need reliable, on-demand translation.

03 — Capabilities

Automation capabilities

Workflow nodes across AI, browser automation, data processing, media, and integrations

Browse all workflow nodes →
Artificial Intelligence
ClaudeGPTGeminiText generationVision AINLPLLM

Build AI workflow automation integrating Anthropic Claude, OpenAI GPT, Google Gemini, and Workers AI for text generation, natural language processing, sentiment analysis, and intelligent automation. Create AI and RAG powered workflows that leverage large language models for content creation, data extraction, translation, summarization, and vision AI capabilities. Automate complex reasoning tasks and build intelligent agents that understand and generate human-like responses.

Browser Automation
Web scrapingScreenshotsPDF generationCloudflareHeadless browser

Automate web scraping, data extraction, screenshot capture, and PDF generation powered by Browser Rendering at the edge. Extract structured data from websites, monitor web pages for changes, capture full-page screenshots, convert web pages to PDF documents, and interact with dynamic JavaScript-heavy sites. Build headless browser automation workflows that run globally with low latency and handle complex web scraping tasks at scale.

Media Generation
Image generationAudio processing3D modelsDocumentsPDF

Generate and manipulate images, audio files, documents, and 3D models in your workflow automation for creative content production. Create AI-generated images, edit and transform existing images with Photon, process and convert audio formats, generate PDF documents from templates, and create 3D models programmatically. Build automated media workflow pipelines for marketing materials, social media content, podcasts, documentation, and digital assets at scale.

Integrations
GitHubDiscordLinkedInRedditGoogle CalendarGmail

Connect your workflow automation with popular platforms including GitHub for repository management, Discord for community engagement and bot interactions, LinkedIn for professional networking and content publishing, Reddit for community monitoring and content posting, Google Calendar for event scheduling and meeting automation, and Gmail for email management and automated communications. Sync data across platforms and build cross-platform workflow integrations.

04 — Open source

Open source workflow platform

MIT licensed and developed in public

Open Source & Self-Hosted

Fully MIT licensed, with no enterprise tier, no fair-code carve-outs, no "community edition" hiding the good parts. Every line of the workflow engine, node library and runtime is on GitHub under the same permissive license. Fork it, self-host it, embed it in a commercial product. Provided as-is without warranty.

Built to Learn AI-Assisted Coding

We started this project to explore AI-assisted development and better understand its capabilities and limits. Development happens in public on GitHub. Contribute custom workflow nodes, report bugs, request features, or observe how we're building serverless workflow automation infrastructure with AI assistance.

05 — FAQ

Frequently asked questions

What people ask before they try Dafthunk

What is Dafthunk?

Dafthunk is an open source visual workflow automation platform. You build workflows by connecting nodes in a React Flow editor, and they run as serverless workflows on Cloudflare Workers and Workflows, with state persisted in D1, R2, and Durable Objects.

Is Dafthunk free and open source?

Yes. Dafthunk is fully MIT licensed, with no enterprise tier, no fair-code restrictions, and no community edition that hides features. The entire workflow engine, node library, and runtime is on GitHub under the same permissive license.

How is Dafthunk different from n8n, Zapier, or Make?

Dafthunk is MIT licensed rather than fair-code like n8n, and it runs natively on Cloudflare's serverless platform. Workflows scale to zero when idle and scale up automatically with demand, with no containers or infrastructure to provision. You can self-host it for free or embed it in commercial products.

Do I need to manage servers or containers to run Dafthunk?

No. Workflows run on Cloudflare Workers and Workflows, so execution is serverless and durable by default. There is nothing to provision or scale. State is persisted in D1 SQL databases, R2 object storage, and Durable Objects, and long-running workflows survive restarts and retries automatically.

What kinds of workflows can I build with Dafthunk?

You can build AI workflows, web scraping and browser automation, data transformations, API integrations, scheduled cron jobs, webhook handlers, and email pipelines using 470+ nodes across AI models, browsers, data processing, media, and third-party integrations.

Can I self-host Dafthunk on my own Cloudflare account?

Yes. The full source is on GitHub and deploys to a standard Cloudflare account using Workers, Workflows, D1, R2, and the Workers AI and Analytics Engine bindings. You own your data, your deployment, and your costs.

Go automate something

Your first workflow takes about four minutes. Your hundredth takes about forty seconds. Cloudflare handles the rest.

Open the editor