Turn your text into a Google Slides deck

The agent works with the Google Slides API v1 and opens a deck by link or by the identifier between /presentation/d/ and /edit: it reads slides, text in shapes and tables, images, page design and thumbnails of individual slides. On that basis it lays your text out across slides — title, key points, conclusions — refreshes metrics and captions across the whole deck at once, and drops data into a ready template without touching its design.

A presentation is built differently from a document: it is not a stream of text but pages of objects, each with its own position and size, and the text itself lives inside shapes and table cells. So replacing something across the whole deck means, for the agent, walking the text shapes of every slide. The look of a page comes from its layout and theme: a slide added from a template layout inherits the design, one assembled out of bare shapes does not.

Hence the caution when writing. The agent adds, duplicates and deletes slides, inserts and replaces text, creates shapes, tables and images, and changes text styles and page backgrounds — but for a reporting deck it prefers to duplicate a template slide and swap the text inside it rather than lay the slide out again by coordinates. Deleting a slide cannot be undone through the API, so the agent confirms it and checks the result against the thumbnail.

Connecting is a sign-in to a Google account in a separate window; access to presentations and to the account address is requested. Google Drive is not part of it, so a deck cannot be found by title — a link or an identifier is needed. Edits travel in a batch: several operations on a deck go out as one request, and if any of them fails none is applied, so the deck is never left half-rebuilt. The agent holds the request rate itself, and states missing permissions or an empty template directly.