The toolbar’s AI button takes an instruction and produces a new node — text or table. You can
ask for multiple outputs, and choose which agent handles it.
The text generate panel rewrites or expands the text in place. Its prompt box explains that
you can connect images, video, audio, and text nodes as references.If you connect a video reference, you are asked to pick an agent with video understanding.
Text nodes also handle documents: import a document to replace the content, and download the
content as Markdown, plain text, or Word.
Add and remove rows and columns, reorder rows, validate column names
AI assistant
Produces a text node from the table
AI table generation
Rewrites the table in place
Field mapping
Choose which column is the text, which is the prompt, and which rows are included
Upload a table file
From a text node’s toolbar, creates a new table node
Field mapping is what makes a table useful: it tells WJI which column to treat as the prompt when
you drag connections out of the table to create one generation node per row.
Paste a YouTube link onto the canvas and a YouTube node appears with the video embedded.Its AI assistant takes an instruction — summarize, rewrite, turn into a script — and produces a text
node named after the video. It requires an agent with video understanding, so the automatic agent is
not offered.
A common workflow: paste a reference video from YouTube, ask the assistant for a shot-by-shot
breakdown, then drag out of the resulting text node to create generation nodes that already carry
the description as context.
Assistant
Responses are generated using AI and may contain mistakes.
