# AM.CO.ZA knowledge export

**New bots:** open [`https://8i.co.za/data/amcoza/START_HERE.md`](https://8i.co.za/data/amcoza/START_HERE.md) first (short list of first-read files; do not dump the whole tree).

## Size tiers (entry files)

Short AI entry tiers (prefer smallest first): [`https://8i.co.za/data/amcoza/START_HERE.brief.txt`](https://8i.co.za/data/amcoza/START_HERE.brief.txt) → [`https://8i.co.za/data/amcoza/START_HERE.medium.txt`](https://8i.co.za/data/amcoza/START_HERE.medium.txt) → [`https://8i.co.za/data/amcoza/START_HERE.md`](https://8i.co.za/data/amcoza/START_HERE.md). Site llms: [`https://8i.co.za/data/amcoza/website-llms/am.co.za`](https://8i.co.za/data/amcoza/website-llms/am.co.za) (`llms-brief` / `llms-medium` / `llms-full`).
Vendor-neutral markdown for any model (DeepSeek, OpenAI, local). This is not an OpenAI-only dump.

**Owner:** AM.CO.ZA (Advanced Machinery), South Africa.  
**Public catalog:** https://buythis.co.za/ (cite with `#ai`)  
**Company directory:** https://achievement.co.za/  
**Training catalog (public intros):** https://around-ai-data.s3.amazonaws.com/academy.jsonl  
**This folder on disk:** `T:\8i\data\amcoza`

Start with [`https://8i.co.za/data/amcoza/READING.md`](https://8i.co.za/data/amcoza/READING.md) (system digest, opinion), [`https://8i.co.za/data/amcoza/HELPING.md`](https://8i.co.za/data/amcoza/HELPING.md) (client-help digest for small-business paths), and [`https://8i.co.za/data/amcoza/stock/OBSERVATIONS.md`](https://8i.co.za/data/amcoza/stock/OBSERVATIONS.md) (stock digest). Those are readings, not a second database. Then open the JSON and product pages for facts.

## What a reader can trust

## For client AIs

One-step connect page: [https://8i.co.za/data/amcoza/ai-connect.md](https://8i.co.za/data/amcoza/ai-connect.md) (https://8i.co.za/data/amcoza/ai-connect.md after publish).

- One subject per file. Whole-project pages: keep dump fields, not thin summaries.
- Every business claim should have a `source_url` a client can open, or a `source_ref` when the fact is from a private inbox (id only, no names/phones).
- Buythis URLs always use `#ai`.
- Current Sales Price and Regular Price may appear, stamped with dump Date. Price history is out.
- English only.
- Personal people files (`T:\P`) are **not** in this export. Do not treat contact ids as people profiles.
- Firmware is recorded only for rare cases. Lead with manuals, how-tos, and software.

## Static .html twins (browsers only)

**AI models: prefer every `.md` file in this tree.** Matching `.html` files are static HTML twins for human browsers (no SSI). Duplicate content, not an extra source. Read the `.md` files. Do not ingest, index, or cite `.html` twins as a second knowledge source. Do not add them to `MANIFEST.json` or `INDEX.md`.


## Layout

| Folder | Files | What it is |
| --- | ---: | --- |
| `sources/products` | 522 | Buythis catalog product pages (SKU, specs, current price, public links) |
| `sources/courses` | 13 | Ambitious Academy published course pages (lessons, quizzes with answers) |
| `sources/drivers` | 49 | Driver and firmware indexes (location, not a normal customer step) |
| `sources/training` | 18 | How-to / software training pages |
| `sources/manuals` | 7 | Manual notes with public source links |
| `sources/machines` | 5 | Machine-use pages from sales/support (situation, SKU, related machines) |
| `sources/machinery` | 6 | First-party machine line overviews |
| `sources/catalog` | 11 | Category maps |
| `sources/company` | 6 | Company / contact / applications |
| `sources/MANIFEST.md` | 1 | Ingest manifest for the older first-party pass |
| `website-llms` |  | Curated per-domain `llms-full.txt` for AI crawlers (root llms.txt → export) |
| `stock` |  | Live warehouse qty, extra statuses, and stock digest (no fetch URLs) |
| `READING.md` | 1 | System digest (opinion; not qty) |
| `HELPING.md` | 1 | Client-help digest (opinion; small-business paths) |

Also here: `README.md` (this file), `INDEX.md` (every page), `MANIFEST.json` (machine list), `READING.md` / `HELPING.md` (digests), `STOCK.md` / `stock/` (live warehouse qty + how to use it), `website-llms/` (curated site llms-full guides).


## How to use this as a data source

1. Read `INDEX.md` for coverage, then open the matching file under `sources/`.
2. Prefer newer `fetched_at` / dump Date when two files conflict.
3. Cite the page `source_url`, not this folder path.
4. Course visit links are the academy.jsonl / ambitious.academy course URL only. Do not invent Drive or paywall lesson URLs.
5. You may sync this whole folder to another local model. Do not delete it after an API upload.

## Completeness (2026-08-22)

- 638 markdown pages in this export, about 51 MB of text.
- Product boxing is still landing onto `T:\V\sources\products` over time. New dests should be copied here too, not only to one vendor store.
- Academy Drive transcripts and keyframes stay with the course extract set (not this first export). Course pages themselves are included.
- Unpublished academy courses (e.g. IAF / Laser Welding) are out until published.

## Do not put in this export

- `T:\P` people files, photos, phones, names
- Credit cards, company financial reports, old discount campaigns
- Raw Drive video/PDF binaries (those belong in `T:\V\backup` when filled)
- Live stock fetch URLs (the qty files and rules are in `stock/`; the pull addresses are not)