Track all your Pancake business conversations in a self-hosted kanban board that shows which partners are going quiet before they churn.
Export a 14-day message trend report as CSV to review your sales team's partner engagement in a weekly meeting.
Use the MCP server to let Claude query your customer database in plain English, such as asking which partners have not replied in two weeks.
Set up a demo of the CRM dashboard to show management how partner health stages work before connecting real data.
Requires a Supabase account (or local Postgres) and the Pancake web interface, Docker Compose simplifies the multi-component setup.
This is an open-source CRM template built for businesses that use Pancake, a customer messaging platform popular in Vietnam that aggregates conversations from Zalo, Facebook Messenger, and Instagram. The project is aimed at Vietnamese B2B sellers and distributors who want to track their partner relationships without storing all that data inside Pancake's own system. The template has four main pieces. A Chrome extension monitors the Pancake web interface and captures conversation data as it appears on screen, then sends it to a small server you run yourself. That server, built with Node.js and Express, writes the data into Supabase, which is a hosted Postgres database. A standalone admin dashboard, packaged as a single HTML file, reads from that database and shows it in your browser. There is also an MCP server, which is a connection layer that lets AI tools like Claude query your customer data using plain-language requests. The dashboard classifies business partners into five stages based on how recently they have been in contact: Active (within 3 days), Sleeping (3 to 7 days), At-Risk (7 to 30 days), Dormant (30 to 90 days), and Churned (over 90 days). These appear as a kanban board. The system also shows a 14-day message trend, a triage view that highlights partners going quiet, and a full partner detail page. CSV export and printable HTML reports are included. Setup can be done in one command using Docker Compose, which starts a local Postgres database, the REST API layer, the sync receiver, and the admin UI together. Alternatively, you can set up each component manually. You connect the admin UI by pasting your Supabase URL and key into a settings screen. A demo mode lets you try the interface without connecting to any database. The project is MIT licensed and built by AIECOS, a Vietnamese company that builds AI tools for local businesses. The README is primarily in English with some Vietnamese. The roadmap mentions future integrations with HubSpot, Salesforce, and Vietnamese commerce platforms.
← aiecosvietnam on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.