NextCRM: An Enterprise CRM for the AI Era

1. Positioning
NextCRM is not a traditional CRM but an AI-native customer management system.
Its core ideas are:
- Let Claude / ChatGPT manage CRM data automatically through an MCP interface
- Use vector search to understand customer intent and history
- Complete enterprise features (invoicing, auditing, multi-currency)
- An API design built for AI agents

This means you can tell Claude: "Find the high-value customers I have not followed up with in the last 30 days."
Claude queries, analyzes and generates a report automatically through the MCP interface, with no manual work.

2. Core features
- AI integration and MCP server
- 127 tool interfaces for AI to call
- Claude / OpenAI API integration
- Vector search and semantic understanding
- E2B sandbox execution
- A complete CRM system
- Account, contact, lead, opportunity and contract management
- Complete sales-process tracking
- Enterprise-grade invoicing
- Multi-currency support
- Tax engine
- PDF generation
- Payment tracking
- Email client
- IMAP/SMTP integration
- Automatic email capture
- Threaded conversations
- Document management
- Upload and version control
- Permission management
- Audit logging
- Advanced search
- Vector database
- Semantic similarity search
- Natural-language queries

3. Tech stack
Layer Technology
Web framework Next.js 16
Frontend React 19 + TypeScript
Auth Better Auth
Database PostgreSQL 17+
ORM Prisma 7.5
Styling Tailwind CSS v4 + shadcn/ui
AI integration Claude / OpenAI API
MCP server 127 tools
Vector search Pgvector + Embedding
Background jobs Inngest
Email service Resend
File storage UploadThing

4. Quick start
# Clone the project
git clone https://github.com/pdovhomilja/nextcrm-app.git
cd nextcrm-app

# Install dependencies
npm install

# Configure environment (API keys required)
cp .env.example .env.local
# Fill in: PostgreSQL URL, OpenAI/Claude API key, UploadThing, etc.

# Run database migrations
npm run db:push

# Start the dev server
npm run dev

# The MCP server can run as a standalone process
npm run mcp:start

5. AI coding friendliness scores

Dimension Score Notes
Directory structure 4/5 Standard Next.js structure, clear MCP integration
Documentation 3/5 Official docs are moderate; MCP interface docs could be more complete
Test coverage 4/5 Has unit tests and E2E tests
Example modules 5/5 Complete CRM, invoicing, email and AI examples
Dependency restraint 2/5 Many dependencies, but all necessary
Incremental dev 5/5 Clear MCP interfaces, easy for AI to generate and extend code

Overall score: 83/100 (best for AI-powered enterprise apps)

6. Why choose NextCRM

1. Currently the most mature AI-native CRM architecture
2. An MCP server with 127 tool interfaces, fully usable by AI
3. Vector search and semantic understanding, beyond traditional CRMs
4. Complete enterprise features (invoicing, multi-currency, auditing)
5. Next.js 16 + React 19, the latest tech stack
6. Can drive business processes directly with Claude or ChatGPT

7. When it is not a good fit

- Many dependencies, a steeper learning curve
- Small teams that need the simplest possible CRM (Atomic CRM fits better)
- Teams on the PHP / Laravel ecosystem
- Traditional CRM needs that do not require AI features