# Desk log KB-SW-FILL-20260822

**Status:** First-party sanitized composite; not a customer extract, independent benchmark, third-party dataset, certification, media report, or endorsement
**Page:** https://infinisynapse.com/en/blog/knowledge-base-software
**Run ID:** `KB-SW-FILL-20260822`
**Date:** 2026-08-22
**Operator:** InfiniSynapse Data Team
**Attestor:** William Zhu, InfiniSynapse cofounder ([GitHub @allwefantasy](https://github.com/allwefantasy))
**Contact for contradictions:** zhuhl@infinisynapse.com

## What this file is

A downloadable record of a wiki fill-rate page versus a bound ops pack on a sanitized shipments extract. It documents the six-step method, input boundary, retrieved-pack counts, and artifact inventory. The companion [aggregate CSV](https://infinisynapse.com/blog-media/knowledge-base-software/downloads/aggregate-KB-SW-FILL-20260822.csv) contains only the published aggregate rows—not the 9,800-row extract. The companion [verify script](https://infinisynapse.com/blog-media/knowledge-base-software/downloads/verify-KB-SW-FILL-20260822.py) checks those published rows only. No independent reproduction was known as of 2026-08-28.

## Six-step method (reproducible)

1. Pick one authorized source. Prefer a replica or sanitized extract.
2. Write one small pack: a field dictionary and one signed report. Do not export the company wiki.
3. Upload the pack, then bind it to that source. Binding is a separate click from upload.
4. Ask one goal with that source selected—not a request for a SQL snippet.
5. Inspect the artifacts and confirm retrieved passages match the signed report, not a wiki draft.
6. Ask the same goal next week. A successful replay cites the same pair.

## Source and goal

| Field | Value |
|---|---|
| Source | Sanitized 9,800-row shipments extract |
| Source grain | One row per sanitized shipment record |
| Source version | Fixed desk extract dated 2026-08-22 |
| Collision | Wiki used shipped ÷ ordered; ops excluded canceled lines |
| Pack metadata | Two-page ops note plus last month’s signed pack; owner: InfiniSynapse Data Team |
| Bind mapping | Ops pack ↔ authorized shipments extract |
| Retrieval configuration | Bound-pack retrieval; configuration values not published |
| Unbound baseline | Same goal with only the wiki page “How we talk about fill rate” |
| Goal asked twice | “What is last-month fill rate?” |

## Grain page (sanitized text)

Fill rate excludes canceled lines. Official name: ops fill rate. Aliases: shipment fill, fill %. Formula: shipped units ÷ (ordered units − canceled units). The wiki sentence “shipped over ordered” is not this rule. Owner: ops. Date signed: last month’s close.

## Retrieved-pack counts

| Retrieval state | Fill-rate rule used | Retrieved notes / reports / exceptions |
|---|---|---|
| Wiki only | shipped ÷ ordered, cancellations ignored | 18 / 10 / 19 |
| Bound two-page ops pack | canceled lines excluded | 16 / 8 / 18 |

Retrieved packs fell because the wiki dump left the bind. Wall-clock for the bound rerun was 11 minutes (warehouse time excluded).

## Task artifacts

| Retrieval state | SQL | Memo | Chart | CSV | Ops pack cited |
|---|---|---|---|---|---|
| Wiki only | 1 | 0 | 0 | 0 | No |
| Bound two-page ops pack | 1 | 1 | 2 | 1 | Yes |

## Minimum disclosure for independent replication

Disclose source grain, row count, and version; wiki baseline; pack versions and approval status; bind mapping; permissions; retrieval configuration; query, SQL, and retrieved passages; all results and failures; run time; and commercial conflicts of interest. Publish confirming and conflicting results.

## What you may cite

- Grain, collision, 18 / 10 / 19 versus 16 / 8 / 18, artifact counts 1 / 1 / 2 / 1, ops pack cited No → Yes, run ID, ~11 min wall-clock

## What you may not claim

- Customer uplift, named-logo case, independent benchmark, certification, raw dataset, official EEAT score, or external validation

## External context (not validation of this run)

- [Stanford HAI AI Index](https://hai.stanford.edu/ai-index) (retrieved 2026-08-28)
- [NIST AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework) (retrieved 2026-08-28)
- [NIST SP 800-53 Rev. 5](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) (retrieved 2026-08-28)
- [CISA Zero Trust Maturity Model](https://www.cisa.gov/resources-tools/resources/zero-trust-maturity-model) (retrieved 2026-08-28)
- [NIST Privacy Framework](https://www.nist.gov/privacy-framework) (retrieved 2026-08-28)
- [OWASP Top 10 for LLM Applications](https://owasp.org/www-project-top-10-for-large-language-model-applications/) (retrieved 2026-08-28)
- [ISO/IEC 27001](https://www.iso.org/isoiec-27001-information-security.html) (retrieved 2026-08-28)
- [ISO/IEC 11179-3](https://www.iso.org/standard/78914.html) (retrieved 2026-08-28)
- [W3C Data Catalog Vocabulary](https://www.w3.org/TR/vocab-dcat-3/) (retrieved 2026-08-28)