WHATELZ.AI
← Blog

2026-04-27

I Built a CRM System Using AI

Not just a pipeline board. The whole chain — capture, manage, follow up, invoice. One lead, one context, one history.

#doublelead#ai#build

What ELZ can you build with AI? How bout a CRM system?

Not just a pipeline board. The whole chain — capture, manage, follow up, invoice. One lead, one context, one history.

That's Double Lead.


The problem it solves.

Most solo operators and small service teams bleed time and money between stranger and payment. They've got one tool for capturing leads, a different one for managing them, something else for follow-ups, and nothing that ties it together. Every handoff is a context switch. Every context switch is a dropped ball.

Double Lead owns the whole chain end-to-end.

What's actually built.

Six modules wired into one product:

  • Lead Generation — embeddable capture forms, WhatsApp/email inbound, lead enrichment via Apollo, Hunter, and Maps. The flagship feature: Lost Leads import analysis, which tells you what went wrong and why.
  • Lead Management — full CRM with a 6-stage kanban, AI-generated summaries and next-action suggestions, team workspaces with role-based write guards, activity feed.
  • Lead Follow-up — Email, WhatsApp, Telegram. Drip sequences. AI auto-reply with timezone-aware active hours. Webhooks.
  • Social Media Management — content calendar, AI caption drafting, asset management through Google Drive, scheduling via Zernio.
  • Customer Management + Invoicing — tuition vertical pre-built, Stripe wired in.
  • Admin — DAU/MAU, activation funnel, deploy info.

The numbers.

115+ API routes. 11 background crons. 35+ database migrations. 295+ unit tests, 24 end-to-end tests. And 5 cloud MCP servers with 43 tools across them — meaning any AI that's connected can read and write to the product directly.

Why that last part matters.

Every competing tool today owns one link in the chain. Their AI is retrofitted on top of something that wasn't designed for it. Double Lead was MCP-native before it had users. The AI isn't a feature — it's the operating layer.


It's live in production. First paying cohort coming in shortly.

Built with a 5-person team. I'm the architect and primary developer.

The moat isn't the feature list — it's the join. One lead shouldn't have to live in four different tools.