Guides, concepts and a full reference of available nodes.
🚧 The documentation is currently under construction. Many things are still missing. But we're working on it...
Learn how to build your first flows
Learn how ro embed the GraVersal chat widget into your website
Learn how to self-host and operate GraVersal
Displays a permanent visible information inside the interactive area of the chat.
Collects data from sub-branches that are thrown in with shovels.
Displays a button that emits a very short signal once it has been clicked.
Send prompts to OpenAI's ChatGPT API and use the response dynamically in your flow.
Clears all messages from the chat interface if reached
Comment nodes are useful if you want to explain things in detail in the flow. Place the node near other structures or connect it to other nodes. This node performs no actions.
This node doesn't do anything special. Use it to forward signals, bundle them, or collect data.
Check if a text contains specific words or sub texts
This node counts how often it has re-entered the context.
Completes the upstream task when reached.
Send emails directly from your flow.
This node executes a series of function calls on other nodes.
Sends or fetches data from an external HTTP API.
Will make the upstream racer the winner of the race if it is reached.
Displays a form in the interactive part of the chat. Requires connected form input nodes and optionally a submit button.
This node imports a subgraph into the running flow.
Connect inputs to form nodes to display input fields
Defines or composes plain JSON data
Displays a button inside the chat that triggers the jump to another flow.
Send prompts to browser native LLM and use the response dynamically in your flow.
A manageable list of data
Fires a message to the chat
Instructs an upstream list node to jump to the next item.
Defines or calculates new numbers
Will be traversed only once per session. Use this node to control idempotence.
Instructs an upstream list node to jump to the previous item.
Displays a print button and generates a print preview from a text template that you can print directly.
Displays a prompt in the interactive part of the chat. Use the entered data directly in the flow.
Displays a question in the chat if in context. Requires connected suggestion nodes, which are displayed as buttons or checkboxes.
A queue node can manage multiple connected task nodes. A queue ensures that all tasks are started cleanly and sequentially.
This is the starting point for a race. Connect Racer nodes to the race node to test different routes simultaneously. The route that reaches the finish node first wins. All other routes will be shut down.
A racer node must be connected to a race.
It reflects the current graph structure along with node states and makes this data directly available to the flow itself.
Takes a list from the flow, generates a button list from it, and makes the selected result available to the flow as data.
It throws data upwards into a bucket where it is collected.
This is the starting node of every flow. Select this node and begin adding your first nodes. Metadata for your flow is also stored here.
Push a state into this node and monitor its changes.
Submits the connected form
Displays answer options in the chat as a checkbox or button next to the corresponding questions.
A switch node can be on or off. Switch this node using call edges or button nodes.
A task must be connected to a queue node. Only one task can run at a time.
Defines or composes texts together
Throw its complete state information back to the upstream try node if triggered.
This node starts a timer when it is activated.
This node is part of the research on business processes in Reactive State Graphs.
This node is part of the research on business processes in Reactive State Graphs.
Allow LLM to use this Tool for their responses
It tests a path and enables a catch path if a throw node is triggered deep in the path.
Tell a tool that the task was finished
This edge is not part of the normal traversal. It just calls state methods on other nodes.
This edge may activate the target node if its optional condition is true when traversed. Optionally it can carry fluid data to the target node.
This edge must be traversed for the target node to activate. If a condition is defined, it must evaluate to true. Optionally it can carry fluid data to the target node.
This edge does not activate the target node on its own. Instead, it prevents activation if it is traversed and its optional condition evaluates to true.
This edge is not part of the normal traversal. It just pushes fluid states to other nodes. The states will disappear if the edge becomes inactive.