← Work
LiveSaaS · CRM · AI Automation2026Social Expert Digital LLC

SED LMS — Lead Management & Operations Platform

A permission-aware business operating system that connects leads, AI website generation, deployment, contracts, mail, payments, tickets and analytics — the whole client journey, one console.

Visit live project
SYSTEM — How it flows
LEADPIPELINEAI SITE GENSTAGINGAPPROVALDEPLOYCONTRACTMAILOPSANALYTICS
LEADPIPELINEAI SITE GENSTAGINGAPPROVALDEPLOYCONTRACTMAILOPSANALYTICS
01 — Overview

SED LMS is the internal operating system I built for Social Expert Digital — a company selling websites, SEO, AI receptionists and custom software to small businesses across the US.

Instead of stitching together a CRM, a site builder, an email client and a contract tool, the team runs the entire client journey from one permission-aware console.

02 — The Problem

Every client moved through the same journey — lead, follow-up, website, review, deployment, contract, payment, support — but each step lived in a different tool with manual copy-paste in between.

Leads slipped through follow-up gaps, website production depended on individual operators, and nobody had one view of where a client actually stood.

03 — Strategy

Model the business process first, then build software around it. The lead record became the single source of truth that every other system feeds from — the website brief, the contract fields, the mailbox thread, the operations tickets.

Automate the repeatable middle (site generation, document assembly, notifications) and keep humans on judgment calls, enforced by explicit review gates.

04 — Architecture

A lead flows through statuses from first contact to closed, with agent and closer assignment, follow-up logging, required next-contact times and overdue/today/upcoming queues.

Lead information becomes the website brief for a five-step operator workflow — setup, content, images, build, review — with AI copywriting, vision-ranked image sourcing and verification gates before anything ships.

One-click staging, client review, redeploys and transfer to the client's own domain close the loop, followed by contract generation from Google Docs templates with placeholder replacement, PDF export and agent signatures.

05 — Development

Built on Next.js 16, React 19 and TypeScript with Supabase for Postgres, auth, storage and realtime — all behind row-level security.

Roughly 90 named permissions are enforced server-side at department and user level, so the same console safely serves agents, closers, operators and admins with dynamically computed access.

06 — Automation & AI

Google Gemini writes site copy from the lead brief; Gemini Vision ranks sourced images before an operator confirms them.

Contracts assemble themselves from templates, the linked mailbox verifies incoming and outgoing delivery, and notifications keep queues moving without anyone chasing status in chat.

07 — Technology
Next.js 16React 19TypeScriptTailwind CSS v4Supabase (Postgres · Auth · Storage · Realtime)Row-level securityGoogle Gemini + Gemini VisionGoogle Drive / Docs APIsIMAP/SMTP (imapflow · nodemailer)ZodVitestHostinger
08 — Results
1Console for the whole client journey
~90Named permissions, server-enforced
5-stepAI website workflow with review gates
v2.13Shipped and iterating in production
09 — Lessons
  • Permissions are a product feature, not plumbing — modeling them early made every later feature safer to ship.
  • AI output needs gates, not vibes: verification steps turned generation from a demo into a production tool.
  • The best automation target is the step people quietly hate — document assembly beat flashier features on impact.
● Available for selected projects

Have a problem worth solving?

Tell me what you're trying to build, automate or improve.