Choose a provider and model, attach context if needed, and send a message.
Open Chat on Web or desktop. Select a configured provider and model, optionally attach files or images, then send your first message. Every user and assistant message becomes a connected node on the canvas.

Keep alternative directions separate instead of flattening them into one transcript.
Start another trunk when you need an independent line of inquiry. To explore an alternative from existing context, continue from any earlier node. The original path remains visible and unchanged.

Use message-level actions without rebuilding the surrounding conversation.
Continue creates a new path from the selected node. Edit revises a message in place. Regenerate asks the selected model for another answer while preserving the conversation structure around it.

Bring exact context from another node into the current prompt.
Type @ in the composer to find a node, or drag a canvas node into the composer. The reference carries that node's context into the current request without merging the histories of both branches.

Keep source material and optional model reasoning near the message that introduced it.
Attach supported files or images before sending a message. When a model returns Reasoning or Thinking content, Flowisall presents it in a collapsible section so the main answer stays readable.

Use built-in providers, custom OpenAI-compatible endpoints, or a local model.
Add your provider credential in Settings, choose a listed model, or enter a custom model name and Base URL. Ollama uses a local compatible endpoint. Provider availability and usage charges remain the responsibility of that provider.

Use summaries and context compression as a conversation grows.
Conversation summaries reduce the amount of earlier material that must be sent again while preserving the visible branch structure. The original messages remain available in local conversation history.

Conversation state and provider settings remain in the current browser profile.
Flowisall stores conversations, canvas state, request logs, and provider settings in the current browser profile. Clearing that profile can remove the data, so export a backup before moving browsers or devices.

Download local conversation data and merge it safely during import.
Data settings can export conversations, canvas state, preserved Agent data, and request logs. API keys are excluded. Import validates the backup and merges records by ID without deleting unrelated local data.

Inspect recent provider, model, duration, status, and token information.
Request Logs groups recent requests by time range and shows the selected provider and model, duration, status, cache information, and input or output token totals where the provider returns them.

Chat is free; desktop project actions require Plus Agent.
Free Chat does not set a conversation-count limit and includes visual branches, edits, regenerations, references, @mentions, and file or image context on Web and desktop. Device storage, model context and provider limits still apply. Plus Agent adds project exploration, planning, file editing, tool execution, Skills, and MCP in the desktop app.

Check the real release endpoint before downloading a platform build.
The Desktop page checks the current release endpoint and reports when no compatible build has been published. Chat is available on desktop; full project actions require an active Plus Agent entitlement.