I've written the LumenDesk 2.0 package: an article for small teams deciding whether to try it now, plus the metadata and editor notes. The article attributes every factual claim to a source file and section, and says Kingy hasn't tested the product. **Length checks** (counted by a script I ran): - **Article:** 778 words, within 600–800. That leaves out the headline and includes subheads and citations, so another counter may differ by a few words. - **Headline:** 61 characters (50–70 required). - **Excerpt:** 37 words (45 maximum). - **Search description:** 159 characters (140–160 required). **How I handled the sources:** - **Price:** the article uses the corrected USD 29 per seat. USD 19 appears only to explain that the launch screenshot was wrong. - **Cost examples:** I worked these out from the pricing file. The three-seat minimum is USD 87 a month, or USD 99 with the scanning (OCR) add-on, before tax. A 750-upload month would add USD 20; that volume is my own example, not a source figure. - **Available now vs. planned:** only existing Team customers in Canada and the US can use it today, in the browser. The API and EU access are planned for Q1 2027, with no date or guarantee. - **Vendor's speed claim** (first drafts down from 50 to 20 minutes): attributed to LumenDesk with its own caveats. It covers drafting only, not checking time or overall time saved. - **Independent review:** described as informal and small. Five of six answers cited the right page and one cited the wrong page, which is too few to give an error rate. - **Internal notes (S5):** none of their claims are in the article. These include "no hallucinations", "GDPR certified", "runs offline", "saves 60%" and the unsourced customer quote. The editor notes give the reason for each. I searched the article for those excluded claims and found none. The only "tested it" match is the sentence saying Kingy hasn't tested the product. I didn't run a style or fact check beyond these. **Open points (also in the editor notes):** - **Needs an editor to confirm:** the sources don't say whether existing Team customers pay extra for the assistant. The article treats Team pricing as the cost and doesn't call it free or an add-on. - **Not in the sources:** when waitlisted customers get access, and whether Team customers outside Canada and the US are included. The article doesn't speculate on either. A human editor would probably need about 20–30 minutes to finish: reread for voice, confirm the pricing question and polish the headline. That's an estimate, not a measurement. Files are in `output/`: - `article.md` - `metadata.json` - `editor-notes.md` - `check.py` and `trim.py` (counting and editing helpers, not for publication)