v2.4 — AI Agent nodes now in beta

Build any workflow.
Visually.

Connect 400+ tools on a drag-and-drop canvas. Watch your data flow in real-time. No code required — but code welcome.

New workflow Watch demo
RUNNING — 340ms
Stripe Webhook
Trigger
payment_intent.succeeded
IF Payment > $100
Conditional
amount_received > 10000
GPT-4o Summarize
AI Agent
Summarize for Slack post
Airtable — CRM Log
Action
Insert row → Deals
Slack #revenue
Output
Post to #revenue-alerts
Ecosystem

Every tool your team already uses

From databases to AI models — each integration becomes a node you can drop onto your canvas and wire into any workflow.

/
S
Slack
Notion
G
Gmail
OpenAI
Airtable
PostgreSQL
St
Stripe
GitHub
Shopify
T
Twilio
400+
integrations and growing
Templates

Start from a real workflow

Production-tested automations built by the community. Clone one, customize it, deploy in minutes.

Payment → CRM → Notify
When Stripe receives a payment over $100, enrich the contact in HubSpot, log to Airtable, and post a summary to Slack.
Revenue Ops
Stripe
→
IF > $100
→
H
HubSpot
→
A
Airtable
→
Slack
Support Ticket → AI Triage → Assign
Incoming Zendesk tickets are analyzed by GPT-4o for priority and category, then auto-routed to the right team in Linear.
Support Ops
Z
Zendesk
→
GPT-4o Triage
→
Route by priority
→
L
Linear
New Lead → Enrich → Score → Sequence
Form submission triggers Clearbit enrichment, an AI-powered lead score, then auto-adds to the right Outreach sequence.
Marketing Ops
Typeform
→
C
Clearbit
→
Score lead
→
O
Outreach
Runtime

Watch your data flow

Every execution is logged node-by-node. Replay any run, inspect payloads, debug in seconds.

execution-log #28491
00:00.0 ▸ Workflow started — Stripe Webhook
00:00.1 ✓ Received payment_intent.succeeded — $247.00
00:00.1 → Evaluating condition: amount > $100 → true
00:00.2 ▸ GPT-4o — generating summary for customer_4891
00:01.4 ✓ GPT-4o response — 142 tokens, 1.2s
00:01.5 ✓ Airtable — row inserted to Deals table
00:01.8 ✓ Slack — posted to #revenue-alerts
00:01.9 ✓ Workflow complete — 5 nodes, 1.9s total
Total executions today
12,847
Peak: 2,100/hr at 14:00 UTC
Median latency
340 ms
P99: 2.1s — includes AI inference
Success rate
99.7%
3 failures in last 24h — all retried
Who builds with Flowline

From ops teams to AI engineers

Different users, same canvas. The visual model adapts to how you think about automation.

01
Revenue Operations
Sync Stripe, HubSpot, and Slack. Automate deal flow, payment reconciliation, and revenue alerts without waiting for engineering.
Stripe → HubSpot → Slack
02
Support Automation
Triage incoming tickets with GPT-4o, auto-assign by priority, close resolved issues. Handle 80% of L1 without human intervention.
Zendesk → GPT-4o → Linear
03
AI Agent Prototyping
Chain language models, tool calls, and memory stores. Prototype multi-step agent pipelines visually before writing code.
Webhook → Claude → PG → API
04
Data Pipeline Orchestration
Extract from APIs and databases, transform with code or AI, load into warehouses. Visual ETL that non-engineers can maintain.
REST → Transform → BigQuery
Open source

Self-host it. Fork it.
Own your automations.

MIT-licensed. Run on your infrastructure. Join 40,000+ contributors building the future of workflow automation.

42k
GitHub stars
1.2k
Contributors
400+
Integrations
terminal
$ docker run -d --name flowline \
  -p 5678:5678 \
  -v flowline_data:/home/node/.flowline \
  flowline/flowline:latest
✓ Container running at localhost:5678