Inspiration
We were inspired by the painful, time-consuming process of turning dense Word docs into presentations. AutoPoint was built to automate the boring stuff and help people present smarter, faster.
What it does
It transforms/translates a Word Document to a PowerPoint and vice versa.
How we built it
We built AutoPoint using Flask for the backend and HTML/CSS/JS for the frontend. We used the python-docx and python-pptx libraries to parse Word documents and generate PowerPoint files. For smart slide summarization, we integrated the OpenAI GPT-3.5 API, which turns each paragraph into a slide with a title and bullet points. The app handles file uploads, processes content in-memory, and returns a downloadable .pptx — all in one seamless workflow.
Challenges we ran into
At first, getting the Word-to-PPT conversion working smoothly was harder than we thought — especially when dealing with different kinds of content like images and weird formatting. We also ran into some API quota issues that forced us to get creative with testing. Setting up CSRF protection and handling file uploads without saving to disk was trickier than expected. And of course, there was the classic "why isn't this CSS centering properly?" moment at 2am.
Accomplishments that we're proud of
We got it working — and honestly, that’s the biggest win. From uploading a Word doc to downloading a clean, AI-powered PowerPoint, everything connects and runs smoothly. Watching a paragraph turn into a real slide felt like magic (after hours of debugging). We're proud we built something actually useful, end to end, in such a short time
What we learned
We learned how to work with file streams in Flask without saving files, how to use python-docx and python-pptx effectively, and how to integrate the OpenAI API for real-time summarization. We also learned the importance of planning the user flow early, managing API keys securely, and that no matter how simple a UI looks, CSS will find a way to mess with you. Most importantly, we learned how to build something fast, functional, and collaborative under pressure.
What's next for AUTO POINT.
Next, we want to make AutoPoint's slides look as good as they work. That means adding professional themes and layouts to each presentation. We're also working on improving how images are handled — ideally placing relevant pictures on the same slide as the text they belong to. Beyond that, we’re exploring customizations like choosing slide formats, fonts, and maybe even AI-generated visuals to match the content.
Log in or sign up for Devpost to join the conversation.