Omni:

Rediscovering the Hidden Assets on Your Computer

Author

Affiliation

Chihiro Kariya, PhD

Kanazawa University

Hidden Assets

Over the years, our computers have become personal archives of knowledge. Research papers, lecture materials, source code, datasets, presentations, and notes accumulate quietly in folders. These files often contain valuable ideas and previous work, but finding them again can be surprisingly difficult. We may remember what we created, but not the exact filename, folder location, or keywords used at the time. This is where semantic search becomes important.

Traditional file search is mainly based on keywords. If a document contains the exact word we enter, it is likely to appear in the results. However, this approach has a limitation: our memory does not always work in keywords. We often remember the meaning of something rather than the exact expression used in a file.

Omni and Semantic Search

Semantic search attempts to overcome this limitation. Instead of simply matching words, it analyzes the meaning of queries and documents. As a result, it can discover connections that ordinary keyword search may miss. Searching becomes less about remembering filenames and more about asking questions about our own information. This is what makes Omni, an AI-powered local search application developed by Han Xiao, particularly interesting. Omni indexes files on your computer and allows you to search your personal archive through meaning rather than only keywords. It aims to help users rediscover information that already exists on their own machines.

After installing Omni, I experimented with several search queries using my own files. In some cases, it successfully surfaced documents that I would not have found through Finder or Spotlight. These were files hidden in old project folders or documents whose contents were relevant even though they did not contain the exact search terms I expected. This experience showed the potential of semantic search as a way to uncover forgotten parts of a personal archive.

As someone who works with both Japanese and English materials, I was also pleased to find that Omni supports multilingual search. Research and teaching materials rarely exist in only one language. I often move between Japanese articles, English papers, code, and presentation materials. The ability to search across languages reflects an important reality of modern knowledge work: our digital archives are increasingly multilingual.

There are still limitations. For example, in my initial experiments, Quarto (.qmd) files did not appear to be indexed. This is an important issue for data analysts and researchers because Quarto documents often combine narrative, code, and results in a single file. I would be excited to see future support for formats like Quarto. Being able to ask a question such as, “Where did I create the figure comparing the debate over Walpole’s removal in the London Magazine and the Gentleman’s Magazine?”, “Where did I create the figure comparing academic performance between high school students who actively participated in inquiry activities and those who did not?” and immediately locate the relevant project would represent a powerful application of AI-powered local search.

Of course, it is too early to know whether Omni will become a widely used application. Search quality, indexing support, and reliability will determine its long-term value. The underlying algorithms are not always visible to users, and I cannot yet say that it will always find exactly what I am looking for. Nevertheless, Omni suggests an important direction for the future of personal computing. The next generation of AI may not only help us create new information; it may also help us rediscover the knowledge we have already produced but forgotten. Our computers are not just collections of files. They are archives of our intellectual activities. Tools like Omni offer a glimpse of a future where these hidden assets become easier to find, understand, and reuse.