Ping

Your relationship memory.

Ping remembers who you met, when to reach out, and what to say — so your network works for you without the anxiety of figuring it all out yourself.

---

What It Is

A native iOS app for students and early-career professionals who meet valuable people but lose touch because they don't know when or what to say.

Core loop: Meet → Log (< 30s) → Get nudged → Edit AI draft → Send

The user always sends manually. Authenticity is everything.

---

Design Docs

All architecture, design, and product decisions live in /design-docs/:

Doc	Contents
01 — Product Vision	What Ping is, who it's for, JTBD, competitive positioning
02 — Architecture	Stack, Supabase schema, data flows, iOS structure
03 — Design System	Colors, typography, components, motion
04 — Navigation & Screens	Screen inventory, wireframes, user flows
05 — AI Pipeline	Embeddings, nudge algorithm, draft generation
06 — Integrations	LinkedIn, Google (Contacts/Calendar/Gmail), Share Sheet
07 — Engineering Setup	Xcode setup, Supabase config, secrets management

---

Stack

• iOS — Swift / SwiftUI, iOS 17+
• Backend — Supabase (Postgres + pgvector + Edge Functions)
• AI — HuggingFace inference (drafts + contact extraction from voice) + Gemini text-embedding-004 (semantic search)
• Auth — Sign in with Apple + Sign in with Google

---

Features

Core Loop

1. Frictionless contact capture — voice (hold-to-speak) or text, under 30 seconds. AI extracts name, company, title, and notes from speech.
2. Semantic search — natural language + goal-triggered contact surfacing using pgvector embeddings.
3. Smart adaptive nudges — AI-scored timing, push notification with draft preview.
4. AI message drafting — tone-calibrated via user writing samples, always user-edited and user-sent.
5. LinkedIn sync — CSV import + iOS Share Sheet extension.
6. Google sync — Contacts import, Calendar scan, Gmail suggestions.

Onboarding

• Guided interactive tutorial — 5-step walkthrough (Welcome → Import → About You → First Capture → Done) that teaches by doing. Users finish with real data in the app. Replayable from Profile → Replay Tutorial.

Voice Input (STT)

• VoiceInputField — reusable hold-to-speak mic component embedded in all freeform text fields:
  • QuickCapture Notes
  • Edit Contact: How You Met + Notes
  • Log Interaction: Note and Interaction Notes
• Streams partial transcripts live; appends to existing text rather than replacing it.
• Uses SFSpeechRecognizer + AVAudioEngine with full permission handling and Settings deep-link fallback.

About You Profile

• Users describe themselves (career role/company/industry/seniority, interests, city, hometown, school, freeform "what I'm looking for").
• Editable from Profile tab → About You, and presented as a skippable tutorial step.
• Feeds into the recommendation engine: CommonalityMatcher surfaces contacts who share interests, industry, school, or company with the user; RecommendationViewModel personalizes the position-tier heuristic based on user seniority.

---

Architecture Notes

Navigation

TabView with 4 tabs: Ping (nudges), Network (contacts), Search (semantic + recommendations), You (profile).

State Management

Modern Swift @Observable macro throughout (not ObservableObject in new code). AuthViewModel is ObservableObject for legacy compatibility with @StateObject / @ObservedObject in the TabView host.

Persistence

• Supabase — all contacts, interactions, nudges, goals, and user profiles.
• Keychain — auth tokens via KeychainHelper.
• @AppStorage — hasCompletedTutorial, tone sample cache flag, UI preferences.

Database Schema (migrations)

Migration	Contents
001	Initial schema (contacts, profiles, interactions, nudges, goals)
002	Enable pgvector extension
003	RLS policies
004	Vector search functions
005	Triggers (updated_at, profile auto-create on signup)
006	Device token for push notifications
007	App config (remote config table)
008	Contact dedup indexes
009	About You profile fields (career, interests, location)