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