Early Access Program
Some features marked with BETA are currently under heavy testing. For early access, contact us at [email protected]
Everything in FlowMason
A complete platform for designing, executing, debugging, and deploying AI-powered pipelines. Core features are production-ready. Advanced features are in beta testing.
Visual Pipeline Builder
DAG canvas editor in VSCode for drag-drop pipeline design with instant validation and live preview.
BetaNatural Language Builder
Describe what you want in plain English. AI generates complete, working pipelines automatically.
Component Library
18+ built-in components: AI nodes (generator, critic, improver), operators (http, transform, filter), control flow.
Pipeline Wizard
Interactive wizard guides you through pipeline creation with templates for common use cases.
BetaAI Copilot
Get AI-powered suggestions, explanations, and optimizations for your pipelines. Generate pipelines from descriptions.
Pipeline Inheritance
Create reusable base pipelines that can be extended. Support for abstract stages, multi-environment configs.
Time Travel Debugging
Step forwards AND backwards through execution history. Inspect data at any point in time.
Prompt Iteration
Edit prompts mid-run and re-execute stages instantly. Compare outputs side-by-side with diff highlighting.
Breakpoints & Stepping
Full Debug Adapter Protocol support. Set breakpoints, step through stages, inspect variables.
Real-time Logs
WebSocket-powered live logging. See execution progress, token usage, and errors as they happen.
Real-time Collaboration
Edit pipelines together with live cursors, presence awareness, and synchronized changes.
BetaPipeline Marketplace
Discover, share, and install pipeline templates from the community. Reviews, collections, versioning.
BetaChat & Comments
In-context communication with team chat, element comments, and threaded replies.
Template Gallery
Curated collection of pipeline templates for common use cases. One-click install and customize.
AI-Powered Analytics
Automatic detection of cost spikes, performance degradation, and optimization opportunities.
Cost Tracking
Track LLM spending by provider, model, and pipeline. Forecast future costs with confidence intervals.
Performance Metrics
P50/P95/P99 latency tracking, slowest stages identification, anomaly detection.
BetaOptimization Recommendations
AI suggests model downgrades, caching opportunities, and reliability improvements with estimated savings.
Multi-Language SDKs
Python, TypeScript, and React SDKs with full streaming support and type safety.
BetaMCP Integration
Connect to AI assistants like Claude via Model Context Protocol. AI-native pipeline management.
Docker Deployment
One-command deployment for development, staging, and production environments.
Webhooks & Scheduling
Event-driven execution with webhook endpoints, cron scheduling, GitHub/Stripe integration.
Control Flow
Conditional branching, foreach loops, error handling with try/catch, routers, and early returns.
Authentication
OAuth 2.0 with PKCE, JWT sessions with rotation, SAML/SSO with signature verification.
Secrets Management
Encrypted storage with AES-256-GCM, rotation support, audit logging, environment isolation.
Rate Limiting
Redis-backed distributed rate limiting. Per-user, per-API-key, and per-endpoint controls.
Audit Logging
Complete audit trail for all operations. Secret access, API calls, pipeline executions tracked.
Three Component Types
Every pipeline is built from these three component categories
AI Nodes
Require LLM providers
generator- Generate text, analyze, summarizecritic- Evaluate and score contentimprover- Refine based on feedbackselector- Choose best from optionssynthesizer- Combine multiple inputs
Operators
Deterministic utilities
http_request- Call external APIsjson_transform- Reshape data with JMESPathfilter- Filter arrays by conditionschema_validate- Validate JSON structurevariable_set- Store values for reuse
Control Flow
Execution orchestration
conditional- If/else branchingrouter- Switch/case multi-routingforeach- Loop over collectionstrycatch- Error handling with fallbacksubpipeline- Call another pipeline
Ready to build?
Get started with FlowMason in under 5 minutes.