DigUp Guide Unofficial
EN

Local AI File Search with DigUp

DigUp provides local AI file search by reading your chosen folders with an on-device embedding model and matching descriptions to pictures, audio clips, and documents.

Current versions: 0.6.1 for Mac, 0.7.0-beta.1 (beta) for Windows and Linux; checked against DigUp's GitHub releases on October 15, 2026.

DigUp runs AI-driven search directly on your computer, indexing only the folders you specify. Rather than relying solely on filenames, it matches the meaning of your query against the contents of images, PDFs, audio recordings, documents, and videos.

Meaning and exact words

When you describe what you need, DigUp checks your query against vectors stored in a local SQLite file. It balances semantic concepts with exact string matches, ensuring terms like invoice numbers or error codes are preserved:

  • “a dog on the beach” locates the relevant photo.
  • “the clause about pets in the lease” opens a PDF directly at that page.
  • “where they talk about sleep” jumps to that minute of an audio track.

Searches can also bridge languages through cross-language search, letting an English query locate Bengali or Arabic documents. For instructions on opening matches or viewing previews, read the usage guide.

Local processing and network activity

DigUp processes files on your hardware. On macOS, it requires an Apple Silicon Mac running macOS 14 or later, which you can install via the Mac download page. 64-bit Windows 10 or 11 and Linux systems are supported through a beta release.

The app only goes online under standard operation:

  1. It downloads the 865 MB model once on first launch.
  2. It checks for application updates once per day, which you can turn off in Settings.

There are no user accounts, no telemetry, and your indexed file content never leaves your machine.

DigUp skips files that live only in the cloud, such as unsynced iCloud or OneDrive files. It also skips keys, certificates, and hidden directories. Code indexing is an opt-in feature kept in a separate index.

Underneath, the search relies on DeepMind’s EmbeddingGemma 2 running through llama.cpp.

Questions

Does DigUp send my file contents over the internet?

No. DigUp reads and indexes your files locally into a SQLite database on your device. The app connects to the internet only to download the 865 MB model file once and to check for daily updates, which can be disabled.

Why are my iCloud or OneDrive files missing from search results?

DigUp skips files that live only in the cloud. Only files downloaded and stored on your local drive are read.

How does DigUp handle exact phrases like invoice numbers?

DigUp pairs 768-dimensional semantic vectors with a keyword index. When you type exact strings such as invoice numbers or error codes, the search gives those matches a boost alongside semantic results.