At Lumona, we're transforming the way people search online.

Searching should be effortless for any question, no matter if you're looking for a website, have a question about what a certain tax form is, or researching the latest methods for fine-tuning LLMs. However, each search option comes with tradeoffs. Some are great at getting you to where you want to go, others are great for answering your questions, and a few are good for research. If you're looking for something on a social media platform, searching is even more difficult, with the search features of these platforms often failing to guide users effectively, leading them to creatively use traditional search engines by adding terms like 'reddit' to their queries to uncover more relevant results. This “hacky” solution only searches Reddit when a more accurate answer or crucial context may lie somewhere else that currently requires additional searches to find. Moreover, the prevalence of misinformation adds another layer of complexity, as unverified data and prioritized SEO-driven results can mislead users.

LLMs have enabled the automatic ingestion and understanding of huge amounts of information. Instead of summarizing links from existing search techniques, our approach is to use LLMs to build our index. This approach enables us to understand each website's meaning, place it into context with similar pages, and fully understand what the user's search query means, allowing us to not only get you where you want to go but answer any question and help you research any topic along the way.

We're a group of 3 MIT students with internships at companies like Google, Stripe, and Reddit who started work on Lumona in September 2023 after winning HackMIT. We have experience planning, building, and shipping new features and projects from start to finish in diverse company environments. Growing up in the age of social media, we are poised to tackle the ever-widening gap between current search solutions and our needs.