AI-Generated Google Slides from Uploaded Documents

Workflow Overview

This workflow automates the creation of professional presentations from user-uploaded files, leveraging AI to extract, summarize, and structure content into Google Slides. The process is interactive, allowing user input and chat history to inform the AI agent’s responses and decisions.

The main goal of the workflow is to scale and automate the task of turning any document into a well-structured Google Slides presentation with minimal manual effort.

Step-by-Step Process

1. User Interaction and File Upload

• Chat Input Node: The workflow begins with a user interface where users can:
  • Enter chat messages, such as instructions or clarifications.
  • Upload files (e.g., documents, PDFs) that will be used as the source for the presentation.

2. File Content Extraction

• File Retriever Node: Uploaded files are processed to extract their content. The workflow can handle various file types, and if needed, apply OCR for scanned or image-based documents.
• Strategy: The extracted content is managed according to a strategy, such as including equal-sized sections from each document, and is limited by a maximum token count to ensure efficient processing.

3. Prompt Construction

• Prompt Template Node: The workflow uses a flexible prompt template:

  • The extracted file content is injected as context ({context}).

  • The user input is included as the current query or instruction ({input}).

  • The template merges these components to generate a prompt for the AI, such as:

    ---DOCUMENT CONTENT---
    {context}
    ---
    ---USER INPUT---
    {input}
    ---
    

4. Chat History Utilization

• Chat History Node: To maintain conversation continuity and context, the AI agent can access the last few (default: 5) chat messages, up to a certain token limit. This helps the agent generate more relevant and coherent responses.

5. AI Language Model Configuration

• OpenAI LLM Node: The system is powered by a configurable OpenAI language model (default: GPT-4.1), with settings for:
  • Model selection (multiple GPT versions available).
  • Token limits (default: 3000 tokens per response).
  • Temperature (controls creativity, default: 0.7).
  • Caching for efficiency.

6. Tooling for Google Slides Automation

• Create Presentation Tool: The agent is equipped to create new presentations in Google Slides, including setting the presentation title, first slide’s title, and subtitle.
• Create Slide Tool: The agent can also add new slides to existing presentations, specifying slide content, title, speaker notes, and optionally, featured images.

7. AI Agent Role and Goal

• AI Agent Node: The central orchestrator is an AI agent with the following configuration:
  • Role: Presentation Specialist.
  • Backstory: 20 years of experience in professional presentations.
  • Goal: “Your task is to create a professional presentation from the content of the file that the user uploaded. You need to go through all of the file and make sure all key information is mentioned in a way in the presentation.”
  • The agent receives the prompt, chat history, file content, and user input, and determines how to structure the presentation.
  • The agent uses the Google Slides tools to create and populate the presentation.

8. Output Presentation & Results

• Chat Output Node: The AI agent’s responses—including confirmation, progress, or presentation links—are displayed back to the user in the chat interface.

How the Components Work Together

Here’s a table summarizing key nodes and their purposes:

Why This Workflow is Useful for Scaling and Automation

• Automates a Labor-Intensive Task: Transforming dense documents into presentations is typically manual and time-consuming. This workflow automates the process, saving hours of work.
• Ensures Consistency and Quality: By using a specialized AI agent with a clear goal and backstory, the output presentations maintain a professional standard.
• Interactive and Adaptive: Users can interact with the agent, provide clarifications, or make requests, and the AI adapts based on chat history.
• Integrates with Google Slides: The direct integration removes the need for manual copy-pasting or designing slides from scratch.
• Scalable for Teams or Multiple Documents: The same setup can be used for batch processing, team workflows, or integration into larger systems.

Example Use Cases

• Corporate Reports: Turn annual reports or meeting notes into ready-to-present slides.
• Educational Content: Convert research papers or textbooks into lecture slides.
• Sales & Marketing: Quickly draft product pitch decks from product specs or whitepapers.

Conclusion

This workflow streamlines and automates the process of generating professional presentations from any document, combining AI-driven content understanding, conversation context, and seamless Google Slides integration. It is ideal for users and teams looking to scale up content creation, ensure quality, and reduce manual labor.