Inspiration
Frustration with the "blank slide" syndrome, Watching hours of productivity vanish while staring at an empty PowerPoint deck, The realization that the most difficult part of a presentation is the initial structure and brainstorming phase. I wanted to create a tool that acts as a cognitive bridge between a rough idea and a professional output, I felt that AI should not just write text but should physically build the tools we use every day. My goal was to empower people who have great ideas but lack the time or design skills to organize them into a logical flow, I was inspired by the possibility of using Google’s Gemini API to automate the architectural work of slide creation.
What it does
Instant generation: It transforms a simple text prompt into a complete and professional PowerPoint presentation in under ten seconds, No more manual formatting or structural brainstorming required. Intelligent Outlining: The app uses the Gemini API to analyze the user's topic and generate a logical sequence of slides, It automatically creates compelling titles and concise bullet points for every section. Native File Export: Unlike other tools that just provide text, this project uses PptxGenJS to build an actual .pptx file, Users can download the file and open it directly in Microsoft PowerPoint or Google Slides for further editing. Modern Web Experience: Built with Next JS and Tailwind CSS, the platform offers a clean and responsive interface that works seamlessly across all devices, It bridges the gap between generative AI and practical document utility.
How we built it
Core Framework: I utilized Next JS for its robust server-side rendering and efficient API routing capabilities. Artificial Intelligence: I integrated the Gemini API to act as the brain of the project, It parses user prompts into structured presentation outlines with logical hierarchies. Document Generation: To convert AI-generated data into a physical file, I used the PptxGenJS library to programmatically define slides and layouts. Styling and UI: The interface was crafted using Tailwind CSS to ensure a modern, clean, and responsive user experience that works across all devices.
Challenges we ran into
Prompt Engineering: Ensuring that the Gemini API consistently returned a strict JSON schema without conversational preamble was a significant hurdle. Data Mapping: I faced difficulties mapping dynamic text lengths into fixed slide dimensions without causing text overflow or layout breaking. State Management: Coordinating the asynchronous flow between the user input, the API response, and the client-side file generation required careful React hook management. File Formatting: Converting the AI's content into a format compatible with the XML-based structure of PowerPoint files required precise implementation of the generation library.
Accomplishments that we're proud of
Seamless Integration: Successfully connecting a high-level LLM like Gemini to a low-level document generator to produce a tangible product. Efficiency: Reducing the time it takes to build a presentation draft from thirty minutes to less than ten seconds. User Experience: Developing a tool that feels intuitive and eliminates the technical barrier between an idea and a professional deck. Robust Logic: Implementing a system that maintains a logical narrative flow across multiple slides regardless of the complexity of the topic.
What we learned
AI Orchestration: I gained deep insights into how to guide Generative AI to produce structured data rather than just creative prose. Full-Stack Development: I strengthened my understanding of Next JS API routes and how to handle heavy client-side file processing. Problem Solving: I learned how to debug complex interactions between third-party APIs and document libraries. Product Design: This project reinforced the importance of building tools that solve a specific, high-friction pain point like the "blank slide" syndrome.
What's next for Serra
Design Customization: I plan to implement themed templates so users can choose the visual style of their generated decks. Multi-Media Integration: Adding the ability to automatically fetch or generate relevant images for each slide using AI. Collaborative Features: Allowing multiple users to edit the same generated outline before exporting to PowerPoint. Advanced Exporting: Expanding the tool to support other formats like PDF or Google Slides directly via cloud integration.
Built With
- gemini
- nextjs


Log in or sign up for Devpost to join the conversation.