Skip to content
BranchesTags

Repository files navigation

🦅 LeadHawk

AI Sales Intelligence Platform

Lead generation, multi-channel outreach, CRM, and analytics — unified with AI.

Next.js TypeScript Tailwind CSS Zustand Supabase Gemini License

Features • Architecture • Quick Start • Tech Stack • Modules • Deploy

🎯 What is LeadHawk?

LeadHawk is a production-grade SaaS platform that consolidates a B2B founder's outbound workflow into one AI-driven cockpit. Instead of juggling LinkedIn, Sales Navigator, email tools, a CRM, and an analytics dashboard, LeadHawk fuses all of it into seven tightly integrated modules powered by Google Gemini (with OpenRouter fallback).

Why it exists: Founders and solo sellers waste hours context-switching between tools. LeadHawk replaces that fragmentation with a single, AI-first surface for prospecting, outreach, social growth, and pipeline management.

✨ Features

🎯 Lead Intelligence

  • AI-generated Sales Navigator filters from plain-English ICP
  • 9-archetype lead briefs (founder, CEO, sales leader, etc.)
  • Personalized hooks + pain points per archetype
  • 4-tone outreach message generator

📱 LinkedIn Growth

  • 6 post types (thought leadership, case study, tips, story, poll, engagement)
  • AI trending topics with 48-hour TTL cache + trend scores
  • 4 / 8 / 12-week growth strategy plans
  • One-click composer pre-fill

🐦 X / Twitter Growth

  • Single tweet + thread + reply generation
  • Full thread builder (hook → setup → insights → CTA)
  • 280-char enforcement with live counter
  • Dedicated Twitter growth plans

💼 Sales Pipeline (CRM)

  • 7-stage Kanban (new → contacted → replied → meeting → proposal → won/lost)
  • Sequence deep linking — auto-trigger email sequences on "contacted"
  • Bulk CSV import with validation + error recovery
  • Daily Action Queue — smart follow-ups + meeting prep
  • AI next-action suggestions

📧 Email Sequences

  • Multi-step sequences (intro → value-add → follow-up → breakup)
  • Auto-triggered when leads move to "contacted"
  • Per-step subject, body, and delay configuration

📊 Analytics & Templates

  • KPIs: Sent, Replies, Reply Rate, Meetings
  • Channel breakdown (LinkedIn / Email / Twitter)
  • Reusable template library with variables {name}, {company}, {role}
  • Response & open-rate tracking per template

🏗️ Architecture

LeadHawk is organized around 7 first-class modules, each with its own page, components, and store slice — all unified by a shared AI router and Zustand state.

graph TB
    User[👤 User] --> Sidebar[🧭 Sidebar Navigation]

    Sidebar --> M1[🎯 Leads]
    Sidebar --> M2[📱 LinkedIn Growth]
    Sidebar --> M3[🐦 X / Twitter Growth]
    Sidebar --> M4[📧 Email Sequences]
    Sidebar --> M5[💼 Pipeline / CRM]
    Sidebar --> M6[📊 Analytics]
    Sidebar --> M7[📚 Templates]

    M1 --> Store[(Zustand Store<br/>+ localStorage)]
    M2 --> Store
    M3 --> Store
    M4 --> Store
    M5 --> Store
    M6 --> Store
    M7 --> Store

    M1 --> AI[🤖 /api/ai]
    M2 --> AI
    M3 --> AI
    M4 --> AI
    M5 --> AI

    AI --> Gemini[Google Gemini 2.5]
    AI -.fallback.-> OpenRouter[OpenRouter Models]

    Store --> LS[(Browser localStorage)]
    User --> Auth[🔐 Supabase Auth]
Loading

Data Flow

Page  →  Component  →  Zustand action  →  localStorage
                  ↘
                    /api/ai  →  Gemini  →  (fallback) OpenRouter

🧩 The 7 Modules

# Module Route Purpose
1 Leads /leads Filter builder + archetype-based lead briefs + outreach generation
2 LinkedIn Growth /linkedin-growth Trending topics + post generation + growth plans
3 X / Twitter Growth /twitter-growth Tweets, threads, and Twitter growth plans
4 Email Sequences /sequences Multi-step cold email sequences with auto-trigger
5 Pipeline /pipeline Kanban CRM, bulk import, daily action queue, sequence linking
6 Analytics /analytics Reply rates, meetings, channel performance
7 Templates /templates Reusable, variable-aware message library

🔧 Tech Stack

Frontend

  • Next.js 14 (Pages Router)
  • React 18 + TypeScript 5.4
  • Tailwind CSS 3.4
  • Radix UI primitives
  • Framer Motion
  • Lucide icons
  • react-hot-toast

State & Data

  • Zustand 4.5 (+ persist)
  • Browser localStorage
  • Supabase (auth)
  • TypeScript-first data models

AI & Tooling

  • Google Gemini 2.5 Flash (primary)
  • OpenRouter (fallback, model cycling)
  • Vitest 4 (unit testing)
  • ESLint + Next lint
  • Custom JSON extractor

🚀 Quick Start

1. Install

git clone <repo-url>
cd leadhawk
npm install

2. Configure environment

Create .env.local:

# Recommended (free tier, generous limits)
GOOGLE_API_KEY=your_gemini_key_here

# Optional fallback (free model cycling)
OPENROUTER_API_KEY=your_openrouter_key_here
OPENROUTER_MODELS=openai/gpt-4o-mini,meta-llama/llama-3-70b-instruct

# Optional Supabase auth
NEXT_PUBLIC_SUPABASE_URL=...
NEXT_PUBLIC_SUPABASE_ANON_KEY=...

🔑 Get keys: Gemini · OpenRouter

3. Run

npm run dev        # → http://localhost:3000
npm run build      # production build
npm start          # serve production build
npm run lint       # eslint
npm test           # vitest

📂 Project Structure

leadhawk/
├── src/
│   ├── pages/
│   │   ├── api/ai.ts              # AI router: Gemini → OpenRouter fallback
│   │   ├── index.tsx              # App shell + page routing
│   │   ├── leads.tsx              # Module 1
│   │   ├── linkedin-growth.tsx    # Module 2
│   │   ├── twitter-growth.tsx     # Module 3
│   │   ├── sequences.tsx          # Module 4
│   │   ├── pipeline.tsx           # Module 5
│   │   ├── analytics.tsx          # Module 6
│   │   └── templates.tsx          # Module 7
│   ├── components/
│   │   ├── layout/                # Sidebar, Header
│   │   ├── leads/                 # FilterBuilder, BriefGenerator, MessageGen
│   │   ├── linkedin/              # PostGenerator, GrowthPlan
│   │   ├── twitter/               # TweetGen, ThreadBuilder, GrowthPlan
│   │   ├── pipeline/              # SequenceDialog, ProgressDisplay, BulkImport, DailyQueue
│   │   └── shared/                # ProfileSetup
│   ├── lib/
│   │   ├── types.ts               # Single source of truth for data shapes
│   │   ├── store.ts               # Zustand global state + persistence
│   │   ├── ai.ts                  # AI generation functions + JSON extractor
│   │   ├── archetypes.ts          # 9 lead archetype profiles
│   │   ├── topicEngine.ts         # Trending topics + 48h TTL
│   │   ├── auth.tsx               # Supabase auth context
│   │   └── supabase.ts            # Supabase client
│   └── styles/globals.css         # Design tokens + utility classes
├── supabase/                      # Supabase config + migrations
├── CLAUDE.md                      # AI development guide
├── PROJECT_OVERVIEW.md            # Full project documentation
└── USER_GUIDE.md                  # End-user manual

🧠 AI Provider Architecture

LeadHawk uses an automatic fallback AI router at src/pages/api/ai.ts:

  1. Google Gemini (primary) — gemini-2.5-flash, JSON mode supported
  2. OpenRouter (fallback) — cycles through configured models on failure
  3. Returns detailed, actionable errors when both fail

All client code calls a single callAI(system, user, opts) interface from src/lib/ai.ts, and a robust extractJSON() helper handles markdown fences, leading prose, and partial wrapping returned by models.

🎨 Design System

  • Theme: Dark-first SaaS aesthetic (#0b1020 base, #1e293b cards)
  • Brand colors: hawk-* (indigo), accent-* (cyan, emerald, amber, rose)
  • Fonts: Syne (display), DM Sans (body), JetBrains Mono (code)
  • Animations: fade-in, slide-up, pulse-slow, shimmer
  • Shadows: glow-indigo, glow-cyan, card-hover

🔑 Environment Variables

Variable Required Purpose
GOOGLE_API_KEY ✅* Google Gemini API key
OPENROUTER_API_KEY ✅* OpenRouter fallback API key
GOOGLE_MODEL — Gemini model ID (default: gemini-2.5-flash)
OPENROUTER_MODELS — Comma-separated OpenRouter model list
OPENROUTER_MAX_TOKENS — Token cap for OpenRouter (default: 800)
NEXT_PUBLIC_SUPABASE_URL — Supabase project URL (auth)
NEXT_PUBLIC_SUPABASE_ANON_KEY — Supabase anon key (auth)
NEXT_PUBLIC_APP_URL — App URL for OpenRouter referer header
NEXT_PUBLIC_APP_NAME — App name for OpenRouter referer header

* At least one AI provider key is required.

🗺️ Roadmap (Build Phases)

Phase Feature Status
1 Lead Intelligence Engine (9 archetypes) ✅
2 Trending Topics Engine (48h TTL) ✅
3 X / Twitter Module (tweets + threads + plans) ✅
4 Analytics Dashboard ✅
5 Pipeline Sequence Deep Linking ✅
6 Money Features (Bulk Import + Daily Queue + Templates) ✅
7 Polish, QA & Type-safety pass ✅
8 Supabase migrations & sync scaffolding ✅

🚀 Deployment

Recommended: Vercel (zero-config for Next.js).

npm install -g vercel
vercel

Then add your environment variables in Vercel → Project → Settings → Environment Variables.

Any host that supports Next.js 14 (Netlify, Render, self-hosted Node) will also work.

⚠️ Known Considerations

  • localStorage limits — large datasets may approach browser quotas; production should sync to Supabase DB
  • AI cost — Gemini free tier can rate-limit; OpenRouter free models also have caps
  • Sales Navigator — requires an active LinkedIn Sales Navigator subscription
  • Trending topics — auto-expire after 48 hours; refresh manually before expiry
  • Sequence trigger — auto-dialog fires on transition to "contacted" only; manual linking always available

See CLAUDE.md for the full list and architectural notes.

📚 Further Reading

Built with Next.js · TypeScript · Tailwind · Gemini

If LeadHawk helped you ship more outbound, drop a ⭐ on the repo.

About

leadhawk-eosin.vercel.app

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages