graph TD %%{init: {'theme': 'mc','layout': 'elk'}}%% ChatInput-wcyi5[<div><img alt="logo" src="/_astro/messages-square.BaSDmT6g.svg" style="height: 20px !important;width: 20px !important"/></div>Chat Input] style ChatInput-wcyi5 stroke:#a170ff googleslidescomponent-0l523[Reemplazar texto] style googleslidescomponent-0l523 stroke:#a170ff ChatOutput-jf3v6[<div><img alt="logo" src="/_astro/messages-square.BaSDmT6g.svg" style="height: 20px !important;width: 20px !important"/></div>Chat Output] style ChatOutput-jf3v6 stroke:#a170ff googleslidescomponent-pmjen[Reemplazar Imagen] style googleslidescomponent-pmjen stroke:#a170ff AdvancedAgent-lja6y[SlideAgent] style AdvancedAgent-lja6y stroke:#a170ff googleslidescomponent-o2b5l[Obtener Presentación] style googleslidescomponent-o2b5l stroke:#a170ff GeminiImageGenerator-9mfx6[Gemini Image Generator] style GeminiImageGenerator-9mfx6 stroke:#a170ff googleslidescomponent-xa0zm[Crear Slides] style googleslidescomponent-xa0zm stroke:#a170ff googleslidescomponent-x60d0[Obtener thumbnail] style googleslidescomponent-x60d0 stroke:#a170ff DeepseekModel-8gvrz[Deepseek] style DeepseekModel-8gvrz stroke:#a170ff googleslidescomponent-pmjen -.- AdvancedAgent-lja6y linkStyle 0 stroke:#a170ff googleslidescomponent-o2b5l -.- AdvancedAgent-lja6y linkStyle 1 stroke:#a170ff GeminiImageGenerator-9mfx6 -.- AdvancedAgent-lja6y linkStyle 2 stroke:#a170ff ChatInput-wcyi5 -.- AdvancedAgent-lja6y linkStyle 3 stroke:#a170ff AdvancedAgent-lja6y -.- ChatOutput-jf3v6 linkStyle 4 stroke:#a170ff googleslidescomponent-0l523 -.- AdvancedAgent-lja6y linkStyle 5 stroke:#a170ff googleslidescomponent-xa0zm -.- AdvancedAgent-lja6y linkStyle 6 stroke:#a170ff googleslidescomponent-x60d0 -.- AdvancedAgent-lja6y linkStyle 7 stroke:#a170ff DeepseekModel-8gvrz -.- AdvancedAgent-lja6y linkStyle 8 stroke:#a170ff

Google Slides Presentation Agent

🧩 Overview

The Google Slides Presentation Agent automatically generates and edits a slide deck from a user’s instruction. By interpreting the request, creating a new presentation, populating text and images, and returning the finished deck, the workflow saves time and ensures consistent formatting across slides.

⚙️ Main Features

Chat‑Driven Interaction – Users send natural‑language requests that the agent understands and acts upon.
AI‑Powered Content Generation – A Deepseek language model produces slide summaries and captions from raw data.
Image Creation – Gemini generates relevant visuals for each section of the presentation.
Google Slides Integration – The agent can create, read, and modify a presentation via dedicated Google Slides tools.
Automated Text & Image Replacement – Specific placeholders are replaced with AI‑generated content and images.
Seamless Output Delivery – The final deck is returned to the user in a single, cohesive response.

🔄 Workflow Steps

Component Name	Role in the Workflow	Key Inputs	Key Outputs
Chat Input	Accepts the user’s request and any attached files.	Text message from the user.	Structured message for the agent.
Slide Agent	Orchestrates the entire process, invoking tools as needed.	User message, configured LLM, list of tools.	Final response message containing the presentation URL or a link to the file.
Deepseek	Generates textual content such as slide titles, bullet points, and conclusions.	Prompt derived from the user request.	Generated text snippets.
Gemini Image Generator	Creates high‑quality images that illustrate slide topics.	Prompt describing the desired visual.	Image files (URLs or binary data).
Crear Slides	Creates a new Google Slides deck based on a template or an empty file.	Presentation title, optional template URL.	URL of the newly created presentation.
Obtener Presentación	Loads the existing presentation to confirm its structure and extract slide IDs.	Presentation URL.	Presentation metadata and slide identifiers.
Obtener thumbnail	Retrieves thumbnail IDs for slide pages to enable image replacement.	Presentation URL, slide page IDs.	Thumbnail identifiers.
Reemplazar texto	Inserts or updates text placeholders within the deck.	Presentation URL, slide ID, search text, replacement text.	Confirmation that text was updated.
Reemplazar Imagen	Replaces existing images on a slide with newly generated visuals.	Presentation URL, slide ID, image URL, image placeholder ID.	Confirmation that the image was replaced.
Chat Output	Formats and delivers the final response to the user.	Response message from the agent.	Chat message displayed in the interface.

🧠 Notes

Authentication – The workflow requires valid Google Slides API credentials and API keys for Deepseek and Gemini; these must be supplied via the tool configuration.
Slide Structure Assumptions – The agent expects a seven‑slide deck with a reserved first (cover) and last (conclusion) slide; other slides contain placeholders for titles, text, and images.
Batch Capability – While the components support batch mode, the typical execution path processes a single presentation per user request.
Rate Limits – Google Slides, Deepseek, and Gemini impose request limits; the agent handles retries internally but may stall if limits are exceeded.
Error Handling – If a tool fails, the agent logs the error and returns a meaningful message to the user rather than failing silently.
Extensibility – Additional tools (e.g., for slide layout adjustment) can be added to the agent’s tool list without modifying the core logic.