How to Make PowerPoints with ChatGPT Using MCP

I have always wondered why native support for PowerPoint generation does not exist directly inside ChatGPT. Sure, Microsoft added Copilot capabilities into PowerPoint itself, but as a user, it would be 100x better if I could simply ask the AI agent to come up with a high-quality presentation without ever leaving the chat interface.

The answer to that missing feature is a Model Context Protocol connector from Plus AI that lets you generate presentation files directly inside ChatGPT. 

Standard ChatGPT can output text outlines or Python scripts to generate basic slides, but the formatting often requires more manual correction than building the deck from scratch. 

This workflow changes entirely when you introduce this protocol. By connecting ChatGPT directly to an external presentation engine, you bridge the gap between text generation and professional slide creation.

What is Plus AI

Before integrating anything into your local environment, we need to look at the engine powering the slides. 

Plus AI provides a dedicated PowerPoint API designed specifically for software developers and automation workflows. You can find the documentation at the Plus AI API features page.

This API does not just dump plain text onto white backgrounds. It applies styling, manages slide layouts, and ensures the output is a native presentation file. When you send a structured payload of headings and bullet points to the endpoint, the API returns a ready-to-present document. This removes the need to manually map AI text into PowerPoint templates line by line.

The service handles the complex XML structure required by modern presentation software. Instead of worrying about font sizes or bounding boxes, you only need to provide the raw content and a chosen design theme.

Understanding MCPs in ChatGPT

The Model Context Protocol acts as a standardized bridge between AI assistants and external tools. 

Originally introduced by Anthropic for Claude, MCP is rapidly becoming the industry standard for giving chatbots secure access to local files, databases, and external web services.

MCP protocol diagram. Image from Model Context Protocol

Instead of pasting text back and forth between ChatGPT and the Plus AI web interface, an MCP server allows ChatGPT to communicate with the Plus AI API directly in the background. You simply ask the assistant to create a presentation. ChatGPT reasons through your request, formats the data exactly as the API requires, and triggers the generation tool autonomously.

The architecture consists of the ChatGPT desktop application acting as the client, a local server running the specific tool logic, and the external API processing the heavy lifting. 

This setup keeps your API keys secure on your local machine while giving the AI model new capabilities.

How to Install the Plus AI MCP in ChatGPT

Setting up this integration requires configuring a local server environment to host the tool. Here are the step by step guide in adding Plus AI in your ChatGPT account as a plugin.

Step 1: Enable Developer Mode

Navigate to the advanced settings page inside your ChatGPT client. Locate the developer mode section and toggle the switch to enable it. 

Enabling developers mode in ChatGPT. Image by Jim Clyde Monge

This exposes the interface options necessary to connect external protocol endpoints.

Step 2: Configure the Plugin

Head over to the plugins page and create a new plugin. Name it “Plus AI” to keep your workspace organized. In the connection field, enter the official endpoint URL: [https://mcp.plusai.com/](https://mcp.plusai.com/). 

Adding a custom plugin in ChatGPT. Image by Jim Clyde Monge

Agree to the terms of service and click the Create button to register the endpoint.

Step 3: Review the Disclaimer

A disclaimer window will appear detailing the permissions the connector requires. Read through the security notices carefully. 

Signing into a plugin server in ChatGPT. Image by Jim Clyde Monge

Once you agree to the conditions, click the “Sign in with Plus AI MCP” button to proceed to the authentication phase.

Step 4: Authenticate Your Account

You will be redirected to a secure login page. This step grants ChatGPT the necessary access to communicate with your Plus AI account. 

Allowing an MCP server to gain access in ChatGPT. Image by Jim Clyde Monge

Click the allow access button to authorize the connection. The system will automatically redirect you back to the ChatGPT interface, and the plugin will be fully active and ready for use.

Generating Presentations with ChatGPT

With the plugin installed, the entire presentation build process happens directly inside your chat window. You no longer need to switch tabs or open separate applications until the PowerPoint file is ready to download.

Prompt the Agent

Open a new chat instance. To invoke the tool, simply mention the plugin by typing @Plus AI MCP followed by your instructions. For my test, I used the following prompt to build a deck for my own publication:

Sample prompt: @Plus AI MCP create a 10 slide powerpoint presentation about the upcoming relaunch of my AI knowledge hub website https://www.zeniteq.com/

Using a custom plugin in ChatGPT. Image by Jim Clyde Monge

Review the Output

The speed and accuracy of the generation process are impressive. After just two minutes of processing, ChatGPT returned a high-quality, well-researched presentation file. 

Using a custom plugin in ChatGPT. Image by Jim Clyde Monge

The visual design matched the aesthetic of Zeniteq perfectly, saving me the trouble of adjusting colors and fonts to fit my brand. More importantly, the contextual information, statistics, and facts laid out across the slides were entirely correct, requiring zero manual factual corrections before presenting.

Save the resulting file to your machine. You can now open it in Microsoft PowerPoint or Google Slides for final visual adjustments and rehearsals.

Why This Workflow Matters

Building presentations usually involves two distinct mindsets. You first need an analytical mindset to structure the argument and gather the facts. 

You then need a design mindset to arrange that information visually on slides. Flipping constantly between these two modes consumes a massive amount of time and often breaks your creative momentum.

Integrating an API through a local protocol delegates both the drafting and the initial formatting to the machine. You retain complete editorial control over the narrative outline, but you completely skip the tedious process of creating text boxes, aligning bullet points, and assigning layout templates.

For developers, solopreneurs, or project managers who need to produce weekly reports or technical course materials, reclaiming that formatting time is a massive productivity gain. 

You can treat presentation decks the same way you treat compiled code. You write the logic in plain text and let the build tool handle the rendering.

Final Thoughts

Alright, that’s about it. We just walked through how the Model Context Protocol bridges the gap between chat interfaces and visual presentation engines. 

By simply enabling developer mode and adding the Plus AI plugin URL directly inside ChatGPT, we essentially taught the assistant how to bypass the tedious formatting phase and generate a finished PowerPoint file from a single prompt.

As a solopreneur who regularly pitches projects and outlines software architectures, I find this workflow incredibly valuable. Instead of breaking my focus to manually format slides and arrange text boxes, I can stay entirely within my development flow. 

I treat the presentation exactly like a software build process where my prompt is the logic and the remote endpoint handles the rendering.

I really appreciate platforms like OpenAI and Plus AI for putting these capabilities directly into the hands of users. Just a few years ago, automating this kind of visual generation required building complex internal tools and managing local server environments from scratch. Having this workflow accessible through a straightforward plugin authentication saves a massive amount of time and mental energy.

I would love to hear your perspective on this setup. Let me know what you think about this workflow and how you plan to use it for your own projects in the comments.