Essence

Takes a single content link and produces structured slides, synchronized voiceovers, and a complete presentation video automatically.

Comment

Inspiration

Creating presentations from existing content is often a repetitive and time-consuming task. Blogs, articles, and videos already contain valuable information, but turning them into structured slides with a clear flow and voice explanation usually requires manual effort. We wanted to reduce this gap between content and presentation by automating the process end to end

What it does

The add-on takes a URL from a blog, website, or YouTube video and automatically extracts the most relevant information from it. This content is then analyzed using a language model to identify key points and organize them into a clear, structured flow suitable for presentation slides.

Based on this structure, the system generates slide-wise content and corresponding voiceovers that explain each slide in context. Finally, the slides and voiceovers are combined to produce an exportable presentation video, allowing users to go from a single link to a complete, narrated presentation with minimal manual effort.

How we built it

We built the backend using FastAPI and handled content extraction and analysis using LangGraph and chat models. The frontend was built with React and Tailwind CSS, and slide generation was handled using the Adobe Document API.

Challenges we ran into

One of the main challenges was managing token usage across multiple chat model calls, especially during slide-wise script generation while preserving enough context for accurate output. Balancing content quality with token limits required several iterations.

We also faced challenges in extracting clean and relevant information from different types of URLs, as blogs, websites, and videos vary significantly in structure and length.

Integrating and registering the add-on with Adobe Express was another hurdle. Understanding the add-on lifecycle and API requirements took time, but Adobe’s documentation and YouTube resources helped us resolve these issues.

Accomplishments that we're proud of

We successfully built an end-to-end workflow that converts raw online content into structured slides, synchronized voiceovers, and a final exportable presentation video within the hackathon timeline. Implementing slide-wise voice generation and seamless video export as part of the Adobe Express add-on was a key achievement for us.

What we learned

This was our first time working with the Adobe Document API, and we learned a lot about building add-ons for Adobe Express. We also gained hands-on experience in structuring LLM workflows and managing model constraints in a real application.

What's next for Essence

We plan to improve the accuracy and consistency of the generated content, add more customization options for slide structure and voice output, and enhance slide design and visual styles. We also aim to support additional content formats and provide more export options for users.

Built With

Share this project:

Updates