Map your professional network. Find your side door.

"The people you barely know are more likely to get you a job than your best friends." — Granovetter, 1973. Confirmed causally on 20 million LinkedIn users in 2022.

🔗 Live Demo →

You've been told to "network." Nobody told you that you already have one.

Even someone with 50 LinkedIn connections is sitting on thousands of shared-context paths they've never mapped. New graduates, immigrants, refugees, and career changers are told to build a network from scratch — but they're not starting from zero. They're nodes in a graph they've never been given a map to read.

Meanwhile, 30–50% of all hires happen through referrals. One referral is statistically equivalent to 40 cold applications. The referral economy is the real labor market. It's just invisible to the people who need it most.

Navox Network makes it visible.

Upload your LinkedIn data export. No login. No server. No data leaves your browser.

🔗 Try it now → or run it locally below.

1. Go to LinkedIn Settings → Data Privacy
2. Click "Get a copy of your data"
3. Select Basic export (ready in ~10 minutes)
4. Download the zip when LinkedIn emails you
5. Drop the zip, folder, or any individual CSV files into the tool — we handle the rest

The app accepts a .zip file, an unzipped folder, or individual CSVs. Connections.csv is the only required file — additional files (messages, endorsements, recommendations, invitations) are used automatically when present to enrich tie strength scoring.

Your data is parsed entirely in your browser using PapaParse and JSZip. Nothing is sent to any server.

git clone https://github.com/navox-labs/network.git
cd network
npm install
npm run dev

Open http://localhost:3000/network and upload your LinkedIn data export.

Requirements: Node.js 18+

After uploading: Zoom into the graph and hover over nodes to see connection details. Check the Outreach Queue tab for who to message first and suggested opener messages by tie type.

This tool is a direct implementation of three bodies of research:

Granovetter (1973, 1983) — The Strength of Weak Ties Weak ties — acquaintances, not close friends — provide non-redundant information about job opportunities because they inhabit different professional clusters. Your close circle recycles the same opportunities. Your peripheral contacts are the bridges.

Rajkumar et al. (2022, Science) — Causal validation at scale A randomized experiment on 20 million LinkedIn users over 5 years, generating 2 billion new connections and 600,000 job placements. Result: weak ties causally increase job mobility. The most valuable connections are moderate-strength weak ties — former colleagues, shared-institution acquaintances — not strangers and not close friends.

Putnam (2000) — Bridging vs. bonding capital Bonding capital (strong ties within a homogeneous group) provides support. Bridging capital (connections that cross group boundaries) provides access to new information, opportunities, and referrals. Job seekers typically have a bonding surplus and a bridging deficit. This tool measures both.

The previous model (recency-only) scored connections by how recently you connected. This is backwards relative to weak-ties theory — a connection from yesterday is likely a stranger; a connection from 3 years ago is an established relationship.

Bridging potential is computed using industry cluster analysis. Each connection is classified into one of 11 clusters: Tech, Finance, Healthcare, Education, Government, Legal, Media/Marketing, Manufacturing, Consulting, Nonprofit, or Other. Connections in clusters with frequency ≤3 in your network are classified as structural bridges per Granovetter — they span otherwise disconnected groups.

Weak ties in bridge positions rank highest in the outreach queue — not strong ties. Strong ties are already activated. Weak bridge ties spanning rare industry clusters are where the undiscovered opportunity lives.

The app ingests a full LinkedIn data export (zip or folder) and automatically detects enrichment files (messages, endorsements, recommendations, invitations) to improve tie strength accuracy beyond what Connections.csv alone provides.

When you upload a full LinkedIn data export, the app uses these files to build a more accurate tie strength model:

Enrichment produces a 3-tier confidence score per connection:

The app supports optional AI features powered by OpenAI or Anthropic.

• Draft Message — generates personalized outreach calibrated to tie strength, network position, and shared context
• Ask Coach — provides contextual network coaching based on your graph analysis

Configuration:

• Add your API key in Settings — the provider is auto-detected from the key prefix (sk- for OpenAI, sk-ant- for Anthropic)
• Keys are stored in your browser only and auto-expire after 30 days
• Connection data sent to the AI provider is disclosed in Settings — review before enabling

No API key is required to use the core network analysis features.

network/
├── app/
│   ├── page.tsx          # Main state orchestration
│   ├── layout.tsx        # Root layout + fonts
│   └── globals.css       # Design system (CSS variables)
├── components/
│   ├── UploadScreen.tsx  # CSV upload, drag-and-drop, export instructions
│   ├── TopBar.tsx        # Navigation + live network stats
│   ├── GraphView.tsx     # Force graph, filtering, node tooltips
│   ├── GapPanel.tsx      # Bridging capital analysis + gap recommendations
│   ├── CompanySearch.tsx # Company/role search with activation guidance
│   └── OutreachQueue.tsx # Ranked targets + draft messages by tie type
└── lib/
    └── tieStrength.ts    # Core model: tie strength, gap analysis, search, ranking

The populations most harmed by AI hiring systems are also the ones least equipped to access the referral economy that bypasses those systems:

• New graduates — rich in peer connections, poor in bridge connections to hiring authority
• New immigrants — existing professional network geographically mislocated; perceived as irrelevant, rarely is
• Refugees — simultaneous credential recognition, employment gap, and name-bias challenges in ATS systems
• Career changers — deep network from previous field treated as useless for new field; almost never true

This tool is built specifically around the insight that these populations are not starting from zero. They are starting from a node in a graph that is larger, more connected, and more navigable than they have been led to believe.

This tool implements findings from:

• Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380.
• Granovetter, M. (1983). The strength of weak ties: A network theory revisited. Sociological Theory, 1, 201–233.
• Rajkumar, K., et al. (2022). A causal test of the strength of weak ties. Science, 377(6612), 1304–1310.
• Putnam, R. D. (2000). Bowling Alone. Simon & Schuster.
• Ryan, L. (2011). Migrants' social networks and weak ties. The Sociological Review, 59(4), 707–724.

Full citation list in the companion research paper: The Invisible Network (Yousif, 2026)

• Navox Coach — AI coaching agent that plugs into your network analysis
• navox.tech — Create a customized AI voice-driven professional profile card to replace your resume — paste the link directly in your outreach messages

MIT — see LICENSE.

Built by Navox Labs · Research by Nahrin Yousif