Local-first career utility

Your career history, ready for your next move.

Keep one verified record of your experience. A required language model runs on your device to compare roles, explain gaps and rank evidence, with no cloud-AI fallback.

Get the latest release Watch the 40-second tour
  • Required local AI
  • No telemetry
  • MIT licensed
CareerOS Local dashboard showing career progress, goals, and application activity
Private vaultStored on this device
Evidence verified
Local vaultYour career facts stay on this computer
Grounded resumesEvery claim starts from a saved fact
Deterministic searchProvider queries use only the settings you enter
Source availableMIT-licensed and inspectable

One record, used throughout

Keep it once.
Use it everywhere.

Save your experience with its evidence, then use the same record across resumes, job review, and application tracking.

Career Vault

Know where every career claim came from.

Organize work, education, skills, and achievements with provenance, verification status, and revision history.

Career Vault displaying verified career facts and their evidence status

Resume Studio

Build from verified experience.

Select saved facts, edit on a visual canvas, and publish immutable PDF or DOCX versions. Nothing is added to your history behind your back.

Resume Studio with fact selection and a visual resume preview

Application Pipeline

Turn discovery into follow-through.

Build provider queries directly from the roles, locations, domains, and limits you enter. Then keep local job snapshots, stages, tasks, and an append-only timeline in one focused pipeline.

Application pipeline showing job opportunities organized by stage

Daily action agenda

Start with the next thing that matters.

One local queue orders overdue, due-today, upcoming and unscheduled actions across every active application. It refreshes when time changes, stays isolated to the signed-in user and never reads private event or dossier bodies.

  • Bounded, deterministic ordering with explicit omitted-item counts
  • Local-day boundaries remain correct through daylight-saving changes
  • Incomplete projections fail closed instead of guessing

Application Readiness

Check the pack before you send it.

CareerOS runs a deterministic preflight over the saved role, application route, profile, linked resume, rendered files and supporting evidence. Every check shows what it found and what to fix. No model is involved.

  • Completeness only, never a hiring prediction
  • The same local state produces the same report
  • Canonical JSON and Markdown exports with a SHA-256 fingerprint

Agent Access

Let an agent inspect the workflow, not take over the vault.

Agent Access uses a separate Python wheel; desktop installers do not add it to your PATH. Build it from a reviewed checkout unless your chosen release lists both the wheel and its hash-locked dependency file. Then create a scoped, expiring grant for Codex, Claude Code or a script. The agent surface offers only approved reads, with no free-form prompt, arbitrary file access or raw SQL.

  • One shown-once token, stored by CareerOS only as a SHA-256 digest
  • Seven bounded bearer reads, registered only when their scopes are granted
  • The desktop app and every agent call share one exclusive vault lease
Install Agent Access

The real app, with fictional data

See the workflow in 40 seconds.

This silent tour uses the working desktop product, not a mock-up of this page. It follows a disposable fictional profile through the dashboard, Career Vault, Resume Studio, and application pipeline.

  1. 00:00 Daily workspace
  2. 00:08 Verified Career Vault
  3. 00:17 Grounded Resume Studio
  4. 00:27 Application pipeline
Reproduce the demo locally

Your data stays with you

Your career is personal data.
Treat it that way.

The API, database, generated documents, and required analysis runtime run on your machine. CareerOS Local adds no telemetry, cloud-model fallback, or remote prompt log.

  • Bounded model context

    Only evidence selected for a task reaches the local model.

  • Explicit network boundary

    Public job searches are separate, visible, user-initiated requests.

  • Portable and erasable

    Export, restore, or explicitly erase the app-owned vault.

Read the privacy model

Your device

YouExplicit actions
CareerOS LocalTauri + loopback API
SQLite vaultCanonical facts
Local modelRequired for analysis
DocumentsLocal artifacts

No product data leaves this boundary

Built for inspection

A desktop application first.
Local AI for every analysis.

Trust boundaries are explicit. Storage is transactional. AI responses follow strict contracts, and Python, React, and Rust checks cover the working system.

Tauri 2 + Rust

Native desktop shell, minimal capabilities, and supervised sidecar lifecycle.

React 19

Keyboard-accessible workspace and an editable visual resume canvas.

FastAPI + Python

Loopback-only application API with validated domain and AI contracts.

SQLite + Alembic

Transactional local state, versioned migrations, and atomic portability.

Continuously verified
Python · React · Rust · migrations · supply chain

View CI history

Run it for yourself

Put your experience to work.
Keep the record on your machine.

Download the release, inspect the source, and try the real workflow with data you control.

Explore the release View on GitHub