FlowDot Web Platform
The FlowDot web platform at flowdot.ai is a comprehensive AI workflow automation system. Build, deploy, and share intelligent automations using a visual node-based editor with 50+ built-in nodes, custom JavaScript nodes, and integrations with major platforms.
TL;DR - Quick Summary
FlowDot Web Platform is your visual AI automation studio:
- Visual Editor: Drag-and-drop node-based workflow builder
- 50+ Node Types: LLM, HTTP, logic, transformers, file processing
- Custom Nodes: Write JavaScript for reusable components
- Apps: Generate React apps from natural language
- Knowledge Base: RAG with semantic search over your documents
- 30+ LLM Providers: OpenAI, Anthropic, Google, local (Ollama)
- Integrations: Discord, Slack, GitHub, Figma, Jira, Trello
- URL: https://flowdot.ai
Core Workflow System
The heart of FlowDot is the visual workflow editor - a node-based graph where you connect inputs, processing, and outputs to create powerful automations.
Visual Workflow Editor
Drag-and-drop interface for building complex automations without code
50+ Node Types
LLM, HTTP requests, conditionals, loops, transformers, file processing, and more
Sync & Async Execution
Instant execution for quick tasks or queued async for long-running workflows
Execution History
Full history with results, timing, token usage, and the ability to continue conversations
Additional Features
- Input Presets (Community Inputs): Pre-configured input values shareable with others
- Shared Execution Results: Share specific outputs with public URLs
- Workflow Validation: Real-time validation of connections and configurations
- Continue Conversations: Resume workflow conversations with full context
/agent - Conversational Interface
The /agent endpoint provides a chat-based interface for interacting with your workflows using natural language.
- Natural Language Discovery: AI-powered intent matching to find and execute relevant workflows
- Chat Interface: Familiar chat UX like other AI platforms
- Multi-workflow Execution: Conversationally trigger multiple workflows in sequence
- Image Attachments: Upload up to 5 files, 10MB each
- Micro-conversation Mode: Condensed responses for edge devices with pre-generated TTS
Memory System
The agent remembers your preferences across conversations:
- Auto-extraction of preferences, facts, and behaviors
- Manual create, edit, and delete
- Toggle individual memories on/off
- Undo auto-generated memories with toast notifications
Agent Toolkits
Bundle related tools together for use by AI agents. Toolkits can wrap HTTP APIs or workflows with credential management.
- HTTP Tools: Define API endpoints with input interpolation
- Workflow Tools: Wrap workflows as callable tools
- Credential Management: Secure mapping of user API keys
- OAuth Integration: Built-in OAuth flows with automatic token refresh
- Toolkit Marketplace: Public sharing with voting, comments, favorites
- Workflow Node Integration: Each tool becomes a draggable node in the editor
Custom Nodes
Create reusable automation components using JavaScript. Custom nodes have typed inputs/outputs and can optionally call LLMs.
- JavaScript Function Editor: Write custom logic in a Monaco-based editor
- Input/Output Schema: JSON Schema for typed sockets
- LLM Capability: Optional
llm.call()for AI-powered nodes - Community Sharing: Publish nodes with voting and comments
// Example custom node
function processData(inputs, properties, llm) {
const result = llm.call({
prompt: `Summarize: ${inputs.Text}`,
temperature: 0.5
});
return {
Summary: result.success ? result.response : result.error
};
}
Apps System
Generate working React applications from natural language descriptions. Apps run in a sandboxed environment and can invoke your workflows.
- LLM-Generated React Apps: Describe what you want, get working code
- Monaco Code Editor: Full IDE experience with syntax highlighting
- Live Preview: Real-time preview pane as you edit
- Workflow Integration: Apps can invoke workflows via
invokeWorkflow() - Multi-file Support: Organize code across multiple files
- App Gallery: Searchable public marketplace
- Categories: Productivity, finance, games, utilities, and more
Knowledge Base (RAG)
Upload documents and query them with semantic search. The knowledge base uses vector embeddings for intelligent retrieval.
- Document Upload: PDF, DOCX, TXT, Markdown, CSV, JSON (up to 50MB)
- URL Import: Download and process documents from URLs
- Automatic Chunking: Smart document splitting for optimal retrieval
- Semantic Search: Vector embeddings for intelligent queries
- Category Organization: Organize documents with categories
- Team Sharing: Share knowledge bases within teams
- Storage Tracking: Monitor usage and limits
Model System
FlowDot supports 30+ LLM providers with a flexible three-tier selection system and BYOK (Bring Your Own Key) support.
Three-Tier Model Selection
- Simple: Fast, cost-effective (gpt-4o-mini, claude-haiku)
- Capable: Balanced performance (gpt-4o, claude-sonnet)
- Complex: Maximum capability (o1, claude-opus)
Provider Options
- FlowDot Credits: Use FlowDot as aggregator with single billing
- BYOK (30+ Providers): Bring your own API keys for direct access
- Audio Providers: TTS/STT with OpenAI, ElevenLabs, Fish Audio, Cartesia
- Local AI (Ollama): Zero cloud dependency option
- Model Browser: Interactive search with capabilities, context length, pricing
Integrations
Connect FlowDot to your existing tools with conversational and event-driven integrations.
Conversational Integrations
- Discord: Slash commands, mentions, button responses
- Slack: DMs, @mentions, slash commands
- MS Teams: Bot messages, mentions
Event-Driven Integrations
- GitHub: push, pull_request, issues, release, workflow_run
- Figma: FILE_UPDATE, FILE_VERSION_UPDATE, LIBRARY_PUBLISH
- Jira: issue_created, issue_updated, comment_added
- Trello: card_created, card_updated, card_moved
Event Mappings
Configure specific events to trigger specific workflows. If no mapping exists, the AI agent decides what action to take.
Research System
Multi-provider web search with automatic fallback and agent auto-chaining.
- Search Providers: DuckDuckGo (free), Google, Bing, Brave, SerpAPI, Serper, Reddit
- Hybrid Execution: Client-side first, server fallback on CORS
- Provider Priority: Configurable priority order
- Auto-Chaining: Agent autonomously chains searches and fetches
Teams & Community
Team Workspaces
- Share workflows, toolkits, and knowledge bases
- 6-level role-based permissions
- Private, public, team, and unlisted visibility options
Community Features
- Public Workflows: Share with the community
- Voting System: Upvote/downvote with quality curation
- Comments: Discussion on all public content
- Favorites: Save content for quick access
- Forking: Copy and customize public content
Scheduled Tasks
Schedule workflow invocations with cron-like scheduling. Runs automatically on the server without CLI.
Environmental Impact
FlowDot is the only platform with built-in energy and CO2 transparency for AI operations.
- Energy Tracking: Per-model energy consumption data
- CO2 Transparency: Emissions data per model
- Environmental Impact Page: Dashboard at /environmental-impact
- Eco-Conscious Model Selection: Choose efficient models