Can AI replace HighLevel?

Build a deliberately narrow local lead-management starter inspired by one slice of HighLevel, in an empty repository.
Do not portray it as a HighLevel replacement.
Use Next.js 15, TypeScript, PostgreSQL, Drizzle ORM, and Docker Compose; do not offer alternative stacks.
The core loop is: capture inquiries, manage contacts and deals, schedule follow-ups, and send a few consented email messages for one business.
Make the first run work locally with one documented command.
Store all data locally by default and provide CSV import and export.
Put secrets in .env, ship .env.example, and never commit credentials.
Create contacts, companies, deals, notes, tasks, configurable pipeline stages, and a next-action dashboard.
Add a public lead form with spam protection, consent capture, validation, and a clear success state.
Add saved filters, search, duplicate detection, merge, and a chronological interaction timeline.
Implement email templates and manual campaign sends through one token-based provider, with unsubscribe, suppression, rate limits, retries, and send logs.
Add role-based access for a tiny team, an immutable activity log, backups, and a recovery command.
Seed a realistic demo workspace and document field mapping for imports.
Include clear empty, loading, success, and recoverable error states.
Write focused tests for the core data flow and one end-to-end happy path.
Create a README with setup, architecture, permissions, data location, backup steps, and sending-domain setup.
Do not add accounts, billing, telemetry, analytics, or a hosted control plane.
Deliberately leave out the operating-system breadth: agency sub-accounts, white-label resale, funnels, websites, memberships, phone and SMS provisioning, payment processing, OAuth integrations, workflow automation, AI, mobile apps, and enterprise reporting.
Do not claim to reproduce managed deliverability, telephony reliability, compliance operations, or a connector ecosystem.
Finish by running the tests and listing the exact commands used.