ProspectKit

A lightweight AI prospecting workspace for founders and small teams.

English · 中文

Build account lists, research companies, find contacts, score leads, and draft outreach without adopting a full sales stack.

Why ProspectKit

Most early teams do not need a heavyweight AI SDR platform on day one.

Apollo, Clay, CRM automations, enrichment workflows, and outbound sequences are powerful, but they can be too much when a founder or a tiny team only needs to answer a few practical sales questions:

• Which companies should we target?
• What does this company do, and why might they care?
• Who should we talk to?
• Is this lead actually a good fit?
• What should the first message say?

ProspectKit focuses on that core loop. It gives small teams one simple workspace for account research and first-touch outbound, with AI built into the flow instead of hidden behind a complex automation builder.

Product Flow

Add companies
    ↓
Generate company research
    ↓
Find relevant contacts
    ↓
Score leads against your ICP
    ↓
Draft personalized outreach
    ↓
Track basic pipeline analytics

Features

Area	What it does
Company workspace	Add target accounts, edit priority, track research status, and manage lists.
AI company research	Uses Exa search plus an AI model to generate structured company reports.
Contact discovery	Finds likely decision makers from public web, LinkedIn, and company pages.
Lead management	Store contacts, update status, set priority, and manage prospect records.
ICP-based scoring	Scores leads across industry, company need, role fit, seniority, and deal size.
Outreach generation	Drafts LinkedIn connection requests, cold emails, or InMail-style messages.
Company Prospecting Brief	One page that orchestrates research, contacts, scoring, and recommendations.
Pipeline analytics	Shows lead volume, A-grade leads, status funnel, score distribution, and trends.
Scoring configuration	Configure your company, product, target industries, target seniority, and weights.

Screens In The App

• /app/dashboard - Discover Companies
• /app/brief/[companyId] - Company Prospecting Brief
• /app/contacts - Manage Leads
• /app/analytics - Pipeline Analytics
• /app/settings/scoring - ICP and scoring configuration

Tech Stack

ProspectKit is built with a modern, deployable TypeScript stack:

• Framework: Next.js 16 App Router
• UI: React 19, Tailwind CSS, shadcn/ui primitives
• Database: PostgreSQL
• ORM: Drizzle ORM
• AI orchestration: Vercel AI SDK
• Search and enrichment: Exa Search API

Quick Start

Requires pnpm (enable with corepack enable if needed).

git clone <your-fork-or-repo-url>
cd prospect-kit
pnpm install

Create .env:

DATABASE_URL="postgres://..."
DATABASE_MIGRATIONS_URL="postgres://..."
EXA_API_KEY="..."

AI_MODEL_PROVIDER="openai"
AI_MODEL_ID="gpt-5.4-mini"
OPENAI_API_KEY="..."

Initialize the database and start the app:

pnpm db:migrate
pnpm db:seed # optional demo companies
pnpm dev

Open the workspace:

http://localhost:3000/app/dashboard

First Run Guide

1. Configure your ICP in /app/settings/scoring.
2. Add target companies in /app/dashboard.
3. Open a company brief and generate research, contacts, scores, and recommendations.
4. Manage contacts and draft outreach in /app/contacts.
5. Review pipeline performance in /app/analytics.

Deploy To Vercel

1. Create a production database

Use any hosted PostgreSQL provider. Neon, Supabase, Vercel Postgres, and standard managed Postgres all work.

If your provider gives you separate URLs, use:

• pooled connection for DATABASE_URL
• direct connection for DATABASE_MIGRATIONS_URL

If it only gives one connection string, use the same value for both.

2. Add environment variables in Vercel

Project Settings -> Environment Variables:

DATABASE_URL="postgres://..."
DATABASE_MIGRATIONS_URL="postgres://..."
EXA_API_KEY="..."
AI_MODEL_PROVIDER="openai"
AI_MODEL_ID="gpt-5.4-mini"
OPENAI_API_KEY="..."

3. Migrate the production database

Before deploying, run migrations against your production database:

pnpm db:migrate

The production build does not automatically create tables. Run migrations explicitly before first use.

4. Deploy

Vercel defaults should work:

• Install Command: pnpm install
• Build Command: pnpm build
• Package Manager: pnpm (lockfile: pnpm-lock.yaml)
• Framework Preset: Next.js

Or deploy with the Vercel CLI:

npx vercel deploy

5. Verify the deployment

After deployment:

1. Open /app/dashboard.
2. Add a company.
3. Generate a company brief.
4. Confirm contacts appear in /app/contacts.
5. Generate a lead score and outreach message.
6. Check /app/analytics.

Commands

pnpm dev          # Run migrations, then start the Next dev server
pnpm build        # Build for production
pnpm start            # Start the production server
pnpm lint         # Run ESLint
pnpm format       # Format the repository
pnpm format:check # Check formatting

Database commands:

pnpm db:auto-migrate # Run migrations automatically
pnpm db:generate     # Generate migration files from schema changes
pnpm db:migrate      # Run migrations manually
pnpm db:push         # Push schema changes directly to DB
pnpm db:studio       # Open Drizzle Studio
pnpm db:check        # Check migration/schema consistency
pnpm db:seed         # Seed demo companies
pnpm db:seed:force   # Force re-seed demo companies

Project Structure

app/(marketing)/page.tsx                 # Public landing page
app/(workspace)/app/dashboard/page.tsx   # Discover Companies
app/(workspace)/app/contacts/page.tsx    # Manage Leads
app/(workspace)/app/analytics/page.tsx   # Pipeline Analytics
app/(workspace)/app/brief/[companyId]    # Company Prospecting Brief
app/(workspace)/app/settings/scoring     # ICP and scoring configuration

app/api/reports/generate/route.ts        # Generate company reports
app/api/contacts/find/route.ts           # Find contacts
app/api/leads/score/route.ts             # Score leads
app/api/messages/generate/route.ts       # Generate outreach copy
app/api/briefs/generate/route.ts         # Orchestrate report/contact/scoring

lib/db/schema                            # Drizzle schema
lib/server/repositories                  # Database access layer
lib/server/services                      # AI, search, and business services
components/workspace                     # Workspace UI
components/shared/ui                     # Shared UI primitives

Security Notes

ProspectKit is designed to be self-hosted and extended. Before exposing it broadly, review these areas for your deployment:

• Authentication and access control
• Team or tenant separation
• API route protection
• Rate limits for AI and Exa calls
• Data retention and deletion policies

For a private internal deployment, you can start with Vercel Deployment Protection, SSO, or a reverse proxy in front of the app.

Roadmap Ideas

• Authentication, teams, and roles
• CSV import and export
• Contact deduplication
• CRM sync with HubSpot, Salesforce, or Attio
• Email sending and sequence support
• Better enrichment provenance
• More customizable scoring rubrics
• Hosted demo workspace

License

ProspectKit is released under the MIT License.

---

中文说明

ProspectKit 是一个面向创始人和小团队的轻量 AI 销售线索工作台。

它可以帮助你建立目标公司列表、研究公司、发现联系人、给线索评分，并生成个性化外联消息，而不需要一开始就引入完整销售系统。

为什么做 ProspectKit

大多数早期团队并不需要从第一天就使用很重的 AI SDR 平台。

Apollo、Clay、CRM 自动化、数据 enrichment workflow 和 outbound sequence 都很强大，但当创始人或小团队只是想验证销售方向时，它们往往太复杂了。早期团队真正需要回答的是几个具体问题：

• 哪些公司值得优先触达？
• 这家公司做什么，可能有什么需求？
• 应该联系谁？
• 这个联系人是不是好线索？
• 第一条消息应该怎么写？

ProspectKit 聚焦这条核心链路。它给小团队一个简单的账户研究和首次触达工作台，把 AI 融进产品流程里，而不是藏在复杂的自动化配置器后面。

产品流程

添加目标公司
    ↓
生成公司研究
    ↓
发现相关联系人
    ↓
按 ICP 给线索评分
    ↓
生成个性化外联消息
    ↓
查看基础 pipeline 数据

核心功能

模块	功能说明
公司工作台	添加目标账户、编辑优先级、跟踪研究状态并管理公司列表。
AI 公司研究	使用 Exa 搜索和 AI 模型生成结构化公司报告。
联系人发现	从公开网页、LinkedIn 和公司页面发现可能的决策人。
线索管理	保存联系人、更新状态、设置优先级并管理 prospect 记录。
ICP 评分	从行业、公司需求、角色匹配、职级和交易规模潜力等维度评分。
外联消息生成	生成 LinkedIn connection request、冷邮件或 InMail 风格消息。
Company Prospecting Brief	在一个页面中编排公司研究、联系人发现、评分和推荐。
Pipeline Analytics	展示线索量、A级线索、状态漏斗、分数分布和趋势。
评分配置	配置你的公司、产品、目标行业、目标职级和评分权重。

应用页面

• /app/dashboard - Discover Companies，公司发现与列表管理
• /app/brief/[companyId] - Company Prospecting Brief，单个公司的销售简报
• /app/contacts - Manage Leads，联系人和线索管理
• /app/analytics - Pipeline Analytics，销售漏斗和线索数据
• /app/settings/scoring - ICP 和评分配置

技术栈

ProspectKit 使用一套现代、容易部署的 TypeScript 技术栈：

• 框架：Next.js 16 App Router
• UI：React 19、Tailwind CSS、shadcn/ui primitives
• 数据库：PostgreSQL
• ORM：Drizzle ORM
• AI 编排：Vercel AI SDK
• 搜索和 enrichment：Exa Search API

快速开始

需要安装 pnpm（可用 corepack enable 启用）。

git clone <your-fork-or-repo-url>
cd prospect-kit
pnpm install

创建 .env：

DATABASE_URL="postgres://..."
DATABASE_MIGRATIONS_URL="postgres://..."
EXA_API_KEY="..."

AI_MODEL_PROVIDER="openai"
AI_MODEL_ID="gpt-5.4-mini"
OPENAI_API_KEY="..."

初始化数据库并启动：

pnpm db:migrate
pnpm db:seed # 可选：插入 demo 公司
pnpm dev

打开工作台：

http://localhost:3000/app/dashboard

pnpm dev 会先运行待执行的数据库迁移，然后启动 Next.js。

第一次使用

1. 在 /app/settings/scoring 配置你的公司、产品、目标行业、目标职级和评分权重。
2. 在 /app/dashboard 添加目标公司。
3. 打开公司 brief，生成公司研究、联系人、评分和推荐。
4. 在 /app/contacts 管理联系人并生成外联消息。
5. 在 /app/analytics 查看 pipeline 数据。

部署到 Vercel

1. 创建生产数据库

准备一个托管 PostgreSQL 数据库。Neon、Supabase、Vercel Postgres 或常规托管 Postgres 都可以。

如果你的数据库服务提供两种连接串，建议：

• DATABASE_URL 使用 pooled connection
• DATABASE_MIGRATIONS_URL 使用 direct connection

如果只提供一个连接串，两个变量可以填同一个值。

2. 在 Vercel 配置环境变量

在 Project Settings -> Environment Variables 中添加：

DATABASE_URL="postgres://..."
DATABASE_MIGRATIONS_URL="postgres://..."
EXA_API_KEY="..."
AI_MODEL_PROVIDER="openai"
AI_MODEL_ID="gpt-5.4-mini"
OPENAI_API_KEY="..."

3. 迁移生产数据库

首次部署前，对生产数据库运行：

pnpm db:migrate

生产构建不会自动创建数据库表。首次使用前请显式运行迁移。

4. 部署

Vercel 默认配置即可：

• Install Command: pnpm install
• Build Command: pnpm build
• Package Manager: pnpm (lockfile: pnpm-lock.yaml)
• Framework Preset: Next.js

也可以使用 Vercel CLI：

npx vercel deploy

5. 部署后验证

部署后建议按顺序检查：

1. 打开 /app/dashboard。
2. 添加一个公司。
3. 生成该公司的 brief。
4. 确认联系人出现在 /app/contacts。
5. 生成线索评分和外联消息。
6. 检查 /app/analytics。

常用命令

pnpm dev          # 运行迁移并启动开发服务器
pnpm build        # 生产构建
pnpm start            # 启动生产服务
pnpm lint         # 运行 ESLint
pnpm format       # 格式化代码
pnpm db:migrate   # 运行数据库迁移
pnpm db:seed      # 插入 demo 公司

数据库命令：

pnpm db:auto-migrate # 自动运行迁移
pnpm db:generate     # 根据 schema 生成 migration 文件
pnpm db:migrate      # 手动运行 migration
pnpm db:push         # 直接将 schema 推到数据库
pnpm db:studio       # 打开 Drizzle Studio
pnpm db:check        # 检查 migration/schema 一致性
pnpm db:seed         # 插入 demo 公司
pnpm db:seed:force   # 强制重新插入 demo 公司

项目结构

app/(marketing)/page.tsx                 # 公开 landing page
app/(workspace)/app/dashboard/page.tsx   # Discover Companies
app/(workspace)/app/contacts/page.tsx    # Manage Leads
app/(workspace)/app/analytics/page.tsx   # Pipeline Analytics
app/(workspace)/app/brief/[companyId]    # Company Prospecting Brief
app/(workspace)/app/settings/scoring     # ICP 和评分配置

app/api/reports/generate/route.ts        # 生成公司报告
app/api/contacts/find/route.ts           # 查找联系人
app/api/leads/score/route.ts             # 线索评分
app/api/messages/generate/route.ts       # 生成外联文案
app/api/briefs/generate/route.ts         # 编排 report/contact/scoring

lib/db/schema                            # Drizzle schema
lib/server/repositories                  # 数据库访问层
lib/server/services                      # AI、搜索和业务服务
components/workspace                     # 工作台 UI
components/shared/ui                     # 共享 UI primitives

安全说明

ProspectKit 适合自部署和二次开发。正式对外开放前，建议检查这些部分：

• 团队或租户隔离
• API route 保护
• AI 和 Exa 调用限流
• 数据保留和删除策略

如果只是内部私有部署，可以先使用 Vercel Deployment Protection、SSO，或在应用前面加反向代理保护。

Roadmap 想法

• CSV 导入和导出
• 联系人去重
• 与 HubSpot、Salesforce 或 Attio 同步
• 邮件发送和 sequence 支持
• 更好的 enrichment 来源展示
• 更可定制的评分规则
• 托管 demo workspace

License

ProspectKit 使用 MIT License 开源。