🧬 Indexa Business OS
The operating system for indexalabs.com — a B2C research-grade peptide shop (Malaysia, RM). This vault is built to be run by an AI agent + you together: every fast-moving thing (an order, a payment, a stock move, a support chat) is captured as a small Markdown log note, and Obsidian Bases turn those notes into live tables and dashboards.
The one-sentence model
Distribution brings demand → Operation fulfils it without breaking → Capital makes sure every cycle leaves more cash than it consumed.
The 3-Pillar Framework
| Pillar | Owns | Question it answers | Folder |
|---|---|---|---|
| 🟣 Distribution | Marketing, content, leads, sales | Where does the next ringgit of revenue come from? | 01 - Distribution |
| 🟠 Operation | Process, workflows, stock, fulfilment, support | Can we deliver every order, on time, in spec? | 02 - Operation |
| 🟢 Capital | Bookkeeping, costing, cashflow, planning | Did the business actually make money — and can it fund the next cycle? | 03 - Capital |
The pillars are a loop, not silos. A single sale touches all three:
TikTok post (Distribution)
→ Lead → Sales Order (Distribution)
→ Stock reserved + Fulfilment + Courier (Operation)
→ Transaction recorded + COGS + margin (Capital)
→ Cash funds next ad / restock (Capital → Distribution/Operation)
How the system is built
Logs, not documents. Anything that happens more than once and has a unique ID (a transaction, an order, a stock movement) is a log note — a tiny Markdown file whose frontmatter properties carry the data. The body is for notes/context.
Bases read the logs. Each .base file filters log notes by tag and renders them as tables, with formulas (ROAS, margin %, available stock) computed live. You never maintain a spreadsheet — you add a note and the Base updates.
One naming + property standard. All IDs, statuses, and property names are defined once in Data Model & Conventions so the Bases, templates, and the agent all speak the same language.
The agent operates it. The Agent Operating Manual tells the AI how to create entries, what to check daily/weekly/monthly, and when to escalate to you.
Start here
- 🎯 Mission & Policies — read first. What we maximise, the non-negotiable rules, the growth-stage cash plan. The constitution.
- 🌳 Value Architecture — Driver Tree — how business value breaks down into the levers an agent can pull (interactive version:
Driver Tree (interactive).html) - 🤖 Agent Operating Manual — daily / weekly / monthly routines the agent runs
- 🗂️ Data Model & Conventions — IDs, statuses, frontmatter schemas, tags
- 📊 KPI Dashboard — the numbers that matter, live from the Bases
- 🟣 Distribution — Strategy
- 🟠 Operation — Playbook
- 🟢 Capital — Playbook
Tech context (current)
Storefront on Vercel + Cursor build, Supabase backend, Lovable for the admin dashboard. Channels: Instagram, Threads, X, Telegram (@indexalabs), TikTok, WhatsApp enquiry, email (nadeem@ / [email protected]). Customer-facing features: product verify, order tracking (→ WhatsApp), COA / test-report pages, membership.
This vault vs. the app
Supabase is the transactional source of truth the website runs on. This vault is the management layer — where you and the agent plan, log decisions, reconcile, and run the business by the numbers. Records here reference app IDs (order #, txn #) so the two stay linked without duplicating the whole database.
Positioning guardrail
Indexa sells research-grade material. All copy, support, and content stays in research-use framing — no human dosing, medical claims, or therapeutic advice. This rule is enforced across Distribution content and Support replies.