Core Features

This is the high-level catalog of what Pathorah does. For deeper dives, follow the links to the dedicated feature pages.

Warm Introduction Pathfinding

The primary feature. Given a source (the logged-in user) and a target (any User, Contact, Company, or Branch), Pathorah finds the N shortest weighted paths through the caller's professional network and ranks them by likeliness. See Pathfinding Engine for the algorithm.

How a Path is Scored

Each edge in a path has a weight on the 0–100 scale. The overall path likeliness is derived from the product of edge weights (via GraphUtils.calculatePathLikeliness), which means a single weak edge tanks the whole path score.

Edge weights by kind:

Edge Kind	Formula	Range
personal-contact	warmth-decay × trust × recency via GraphUtils	0–100
employment	base (current/past tenure) + seniority lift + admin bonus, capped at 90 for unverified	0–90 (regular), 95–100 (verified rep)
company-rel	type base × strength × recency multiplier	0–100
knowledge	raw weight × confidence	0–100

Target Types

POST /api/paths/find accepts:

• targetUserId / targetContactId — legacy person targets
• targetType + targetId — polymorphic targets including Company and Branch
• allowTransitNodeTypes — which node types may appear as intermediate hops

Introduction Requests

Once a path is found, the user can send an introduction request along the chain. Each intermediary is notified via WhatsApp (Omevision shared notification service) and can accept or decline.

Contact Management

Users maintain an address book of Contacts — people who are not yet (or may never be) Pathorah users. Contacts participate in the graph exactly like Users; the only difference is they cannot log in. Employment, personal connections, and all graph edges apply equally. The userLinkAdapter bridges a Contact to its corresponding User node when the contact eventually signs up.

The Network

The full conceptual model — personal plane vs. knowledge plane, ownership scope, edge kinds, growth loops — lives in The Network. Skim it before diving into pathfinding internals.

Companies & Branches

Companies are first-class peer nodes since 2026-04-07. Users can declare employment at any company (creating it on the fly via free-text), companies can have branches with their own employee lists, and the eight company-to-company relationship types let pathfinding hop through partner / supplier / acquired / parent links. See Companies & Branches for the full user-facing guide.

Representative Verification

Anyone can claim "I am the founder/principal of this company," but only verified claims earn structural-dominance scoring in pathfinding. The three-tier ladder (self-declared → email-verified → document-verified or vouched) is documented in Representative Verification.

Knowledge Graph & AI Enrichment

Companies are tagged with industries, capabilities, products, markets, and certifications drawn from canonical taxonomies. The AI service (Gemini-backed) can suggest these and propose company-to-company relationships, with all suggestions surfacing as drafts that an admin reviews before they become real graph edges.

Circles

Users group their personal contacts into Circles (Family, Investors, Eswatini Tech, etc.). Circles drive pathfinding scoping ("only paths through my Investors circle"), permission-style features for posts, and future scoped intro templates.

Credits & Subscriptions

Path searches consume credits. Credit gates are enforced in middleware before the BFS engine runs. Subscription tiers control the monthly credit replenishment rate and the maximum path depth.