On-device search, routing, and semantic recall

Everything on your Mac,
instantly within reach.

Vector brings apps, files, messages, clipboard history, weather, maps, contacts, and calendar into one launcher. It feels instantaneous because the intelligence stays local.

Download Vector
Search across: Apps Files Messages Clipboard Calendar
Instant app launching Semantic file search Message recall Clipboard history Maps and places Calendar and weather Local ML routing Liquid Glass interface Instant app launching Semantic file search Message recall Clipboard history Maps and places Calendar and weather Local ML routing Liquid Glass interface
Features

A launcher with
actual context.

Vector is opinionated about what should happen when you type. It ranks for intent, not just string matches.

Apps First

Open Vector, type a few letters, and launch. Keyboard-first ranking keeps app results immediate and predictable.

Zero hunting

Routing Model

Type a person, get contacts. Type an address, get maps. Vector routes the query before you have to think about sources.

Intent-aware

Semantic Search

Ask natural questions about files and conversations. Vector searches for meaning instead of just matching tokens.

Meaning over keywords

Clipboard History

Copied something hours ago? Search it instantly from the same panel without switching mental modes or opening another utility.

Always recallable

Ambient Context

Calendar in the morning, weather before leaving, place results when you need directions. The panel adapts to your day.

Contextual surfaces

Beautiful By Default

Vector leans into the latest macOS materials, smooth motion, and customizable layout so it feels native instead of bolted on.

Liquid Glass ready
Ambient Intelligence

Useful before you finish the thought.

Vector can surface the right utility panel based on time, context, and intent, so the launcher starts feeling more like a live command surface.

  • Morning agenda and travel context without opening Calendar.
  • Weather, maps, and contacts in the same search workflow.
  • A single interaction model instead of five tiny menu bar tools.
Vector showing contextual calendar and weather results
Semantic Search

Files and messages,
searched by meaning.

Natural language queries work because embeddings and retrieval stay on-device. You can search personal context without handing it to a server.

  • "When was Joseph planning on visiting?"
  • "Which Swift file manages networking?"
  • "That note about the airport pickup."
Vector semantic search showing iMessage results
Tailored Workflow

Tune the launcher to your habits.

Move the panel, change sources, control shortcuts, and decide how much Vector should know about your system. The customization is deep without being noisy.

  • Adjust panel placement for your desktop geometry.
  • Configure which sources show up and when.
  • Keep the app feeling personal instead of universalized.
Vector settings panel showing customization options
Privacy

Private by design,
not by marketing.

Vector’s strongest advantage is architectural. The intelligence runs locally, which means low latency and a saner trust model.

On-Device Processing

Embeddings, ranking, and query understanding run on your Mac. Searches do not need to leave the machine to be useful.

Works Offline

Semantic search and intelligent routing still work without a cloud dependency, which also means lower latency and fewer failure modes.

Optional Analytics

Anonymous TelemetryDeck analytics can help improve the product, but opting out is simple and no personal content is collected.

Download

Pay what you want.
Keep it forever.

Vector is built to replace the friction of Spotlight without adding subscription fatigue. Download it, tune it, and make it yours.

Download for macOS

Apple Silicon · macOS 26+ · No recurring fee

FAQ

Questions people
actually ask.

It's around ~120MB right now. Vector bundles two ML models: a routing model that decides which sources to surface for a query, and an embedding model for semantic search across messages and files.

Vector uses the Neural Engine where possible and aggressively optimizes memory. Semantic indexing can temporarily use up to around 800MB while it runs, but idle memory is usually far lower and still being improved.

Quit Vector and move the app to the Trash. You may also want to remove ~/Library/Application Support/Vector afterward because indexing data can take a few hundred megabytes.

You need an Apple Silicon Mac running at least macOS 26.0, plus enough local storage for indexing. Performance is best on newer chips.

Major issues are typically addressed quickly, and new features land whenever they are ready enough to feel stable rather than rushed.

You can, but Vector is intended more as a superior Spotlight replacement than a Raycast clone. It is optimized for people who want search to feel native, fast, and reliable again.

Fried rice still leads, especially if it has the same energy as the best Chinatown plate ever. Steak, shawarma, and a good milkshake are also difficult to argue with.