Product-grade AI systems

A focused portfolio of work in logistics, content, lead handling, and document intelligence.

These are the systems we put in front of clients because they show the pattern clearly: AI inside the workflow, useful in production, and built to keep working when real teams depend on them.

Logistics intelligence system
AI system 01

Logistics intelligence system

Keeps shipment updates current and exception handling organized across carrier inboxes.

  • What it does: Reads carrier messages, detects operational changes, and syncs updates into the logistics workspace.
  • Why it matters: Gives operations teams a live view of exceptions without manual inbox work.
  • Stack: React, Supabase, Node.js, and automation.
AI React Supabase Automation Node.js
Content operations copilot
AI system 02

Content operations copilot

Turns a single brief into channel-ready content variations for web and social.

  • What it does: Drafts posts, carousels, and versioned copy from one prompt.
  • Why it matters: Keeps planning, generation, and review in one workflow.
  • Stack: React, Claude, Node.js, and automation.
React Automation Claude Node.js
Inbound lead qualification system
AI system 03

Inbound lead qualification system

Qualifies website leads, collects consent, and prepares a concise handoff brief only when the fit is real.

  • What it does: Captures contact details, project context, and consent in chat.
  • Why it matters: Researches company and domain signals before the handoff.
  • Stack: Python, Claude, SQLite, Telegram, and ERPNext.
Python Claude SQLite Telegram ERPNext
Document intelligence workspace
AI system 04

Document intelligence workspace

Gives teams paragraph- and cell-level control inside Google Docs, Sheets, and Slides.

  • What it does: Handles document, spreadsheet, and presentation operations.
  • Why it matters: Reduces manual editing across Google Workspace.
  • Stack: TypeScript, MCP, and Google API.
TypeScript MCP Google API

Supporting systems

Useful builds we keep public without putting them ahead of the core AI work

Browser automation layer

Controlled Chromium sessions for repeatable research, monitoring, and browser workflows.

Docker REST API Playwright

Shift coordination bot

Telegram workflow for attendance proof, task planning, and automated reports.

Python aiogram PostgreSQL

Meeting stack

Self-hosted scheduling, video conferencing, and AI transcription for internal calls.

Docker Compose Jitsi Cal.com

Public-sector workflow audit system

Structured document collection and workflow analysis for audit and reporting work.

Python Google Drive LLM