How a page gets published on weaver.work

Weaver content is researched, drafted, and reviewed by named human authors. AI tools (specifically Anthropic Claude and OpenAI ChatGPT) may be used as drafting and editing assistants — for research synthesis, sentence-level rewriting, outline generation, and reviewing a draft against an editorial rubric.

Final accountability for accuracy, expertise, and editorial judgment rests with the named author of each page. No page on this site is authored solely by AI without human review. When AI is the primary drafter on a specific page (rather than an assistant on prose the author already wrote), that page carries a per-page disclosure in the body or footer naming the tool, the human reviewer, and the review date.

AI-generated images are not used as hero or section assets on editorial pages. If that ever changes, the affected page will disclose the generator (Midjourney, Imagen, DALL-E, etc.) and the relevant Content Credentials metadata in the image caption.

Quantitative claims on this site fall into three categories and are sourced accordingly:

• Industry / academic claims. Cited inline with a hyperlink to the primary source (peer-reviewed paper, government / standards-body publication, vendor whitepaper). The same sources are consolidated on /research.
• Customer outcomes. Customer-named numbers (revenue, time-saved, headcount- affected) come from the customer’s own attestation and are reproduced verbatim with the customer’s permission. Anonymized numbers do not appear on this site.
• Internal benchmarks. Performance numbers about the Weaver platform itself (latency, query times, throughput) are reproducible on a documented runtime / dataset / configuration. The methodology lives next to the claim — either inline on the page or on /trust/methodology when that standing page ships.

There is no number on this site we are not prepared to walk a reader through end-to-end. If you find a claim with a missing or rotted source, write to [email protected] and we will publish a correction or fix the citation.

Pages on this site fall into two classes:

• Original research. The page presents data the author collected, an analysis the author performed, or a framework the author defined. Charts, tables, and screenshots come from the author’s own work, not third-party sources.
• Synthesis or summary. The page summarizes a body of external research or industry practice for a non-specialist reader. Such pages are labeled in the introduction and lean entirely on the citations in §3 above.

We do not publish “original research” that is actually a re-write of someone else’s research with the citation removed. If the work is a synthesis, the page says so.

When a fact, number, or attribution on this site is wrong, we publish a correction under the same byline rather than silently editing the page. The correction names the original claim, the corrected claim, the date of the correction, and (when relevant) why the original was wrong.

A page’s “last reviewed” date reflects an editorial change — new content, new analysis, a correction, a substantive rewrite. It does not reflect a technical refactor like a rendering or schema-only fix. We do not move the review date forward unless the editorial content actually moved.

If a claim turns out to be unsalvageable, we retract the claim on the same page rather than removing the page. Removed URLs become 404 referrers in someone else’s analysis; retracted claims become public corrections.

Weaver authors are part of the team building the Weaver platform. That is a structural conflict of interest, not a hidden one: every comparison page, every benchmark, and every customer story is published by someone with a commercial interest in the outcome. We disclose that openly here rather than pretend otherwise.

What that disclosure obligates us to:

• When a comparison page names a competitor, we link to the competitor’s own site so the reader can check our characterization against theirs.
• When a benchmark favors Weaver, the methodology is on the same page so a reader can reproduce or contest it.
• When a paid customer story uses a customer’s name and numbers, the customer authorized that use in writing. The authorization is on file and available on request.

If you find an error of fact, a missing or rotted citation, a conflict between something on this site and an author’s external profile, or any other editorial issue, write to [email protected].

We commit to acknowledging the report within one business day and either publishing a correction, publishing a clarifying note, or explaining why the original claim stands. Either way the response goes on the page, not into a private email thread.

Security disclosures use a different channel — see the Hikma security disclosure page (in development) or contact [email protected].