The agent is the graph you build.
PretzelGraph is a visual agent runtime. Design how your agent thinks. Its loop, its tools, its memory, all on one canvas. Then run it on its own, or drop it in as a building block inside a bigger workflow.
Free workspace on its own private machine · open core · no card required
Connect your stack
OpenAI
Anthropic
Gemini
Grok
OpenRouter
Tavily
MongoDB
PolymarketDraw the loop. Ship the agent.
From an empty canvas to a running agent other people can embed.
Trace and debug your graphs
Every run is recorded as a timeline. Open any node to see what went in, what came out, how long it took, and what it cost in tokens. When an agent does something odd, scroll back to the exact step and find out why.
Your agents get their own machine.
The isolation big platforms sell as an enterprise add-on is just how PretzelGraph works — and the whole product self-hosts with one Docker command, free.
Privacy and isolation
Every cloud workspace runs on its own virtual machine with its own database. Prefer your own hardware? Self-host the identical product, free.
Publish what you build
List a workflow publicly and anyone can drop it into their canvas as a single node. When you publish an update, everyone who embedded it receives it.
Pricing that scales with you
Start free. Upgrade when your agents need to stay awake and run more.
Free
Explore the features and get started.
- No credit card required
- 100 executions a month
- 2 concurrent executions
- 1 day of execution history
- Sandboxed arbitrary code execution
ProPopular
For your first agents in production.
- 2,500 executions a month
- 5 concurrent executions
- 7 days of execution history
- Sandboxed arbitrary code execution
Pro+
For heavier workloads and longer histories.
- 10,000 executions a month
- 10 concurrent executions
- 30 days of execution history
- Sandboxed arbitrary code execution
Frequently Asked Questions
How PretzelGraph runs, where it runs, and what you keep. If something is not answered here, get in touch.
