Editorial Policy
Editorial, Sourcing and AI-Use Policy
What governs the content on this site?
MartinZialcita.com publishes original analysis, frameworks, and documented project work by Martin Zialcita. Every factual claim is classified in a public register before publication. AI tools assist with drafting and editing but do not generate claims, evidence, or conclusions. Sourcing standards, correction procedures, and review cadences are documented here and applied consistently across all pages.
54-word direct answer
Purpose of this policy
This policy describes how content on MartinZialcita.com is produced, sourced, reviewed, and corrected. It applies to all pages, articles, frameworks, and research published under the martinzialcita.com domain.
The site exists to share Martin Zialcita's practical knowledge of AI implementation, systems integration, and workforce readiness. Content standards are designed to support accurate retrieval by search engines and AI answer systems, not merely to satisfy legal requirements.
First-hand experience versus third-party evidence
Much of the content on this site draws on Martin's direct, first-hand involvement in the work described. First-hand evidence means Martin personally designed, built, deployed, trained for, or observed the systems and engagements he writes about. When that is the basis for a claim, the page says so explicitly.
Third-party evidence (client testimonials, external research, published data) is handled differently. Client outcomes are described only in approved wording and only with documented permission. Published research is linked to its primary source, not paraphrased from secondary summaries. Where a claim is attributable to a third party's observation rather than Martin's own measurement, that attribution is stated.
The distinction matters because first-hand experience and third-party evidence carry different epistemic weight. A testimonial describes what a client perceived; a measured benchmark describes what was counted under defined conditions. This site does not conflate the two.
AI assistance disclosure
AI tools, including large language models, are used in the production of this site. Their role is limited to drafting assistance, copy editing, structural suggestions, and code generation. AI tools do not generate claims, evidence, outcomes, client descriptions, or conclusions. All substantive assertions originate with Martin or with documented sources.
Where an article or page was substantially drafted with AI assistance, that is noted in the article's metadata or footer. AI-assisted copy is reviewed, edited, and approved by Martin before publication. The presence of AI assistance does not reduce the sourcing standard; every claim must still meet the criteria described in this policy.
Martin does not use AI to manufacture quotations attributed to himself. Labeled principles, definitions, and takeaways on this site are his actual positions, reviewed before publication.
Claim classification and the public register
Every important factual claim is classified before it reaches a published page. The classification system has four statuses: Verified, Qualified, Unverified, and Retired. Only Verified and Qualified claims may appear on published pages, in their approved wording.
Aggregate performance statistics (percentages, dollar figures, efficiency multipliers) are withheld until the underlying evidence, measurement method, and client permissions are documented. This is not false modesty; it is how claims earn the right to be published.
Where to read the full claim standard
The four-status model, all eleven register fields, and the full list of claims and their current status are published at /claim-policy/. That page is what the homepage links to when it explains why no performance percentages appear.
Sourcing standard
Before a factual claim appears on a published page, it must satisfy all of the following.
Origin identified
The claim traces to a primary source, first-hand observation, or named client testimonial, not to a secondary summary or another website that may have its own sourcing gaps.
Approved wording used
The exact wording of the claim matches the approved wording in the claim register. Paraphrasing a verified claim into something stronger is prohibited.
Permission documented for attributed claims
Client names, testimonials, and project descriptions are published only with documented permission from the relevant person or organization.
Primary source linked
External research and data are linked to the publishing institution, not to a news article that mentions the research.
Limitations stated
Where evidence has a scope boundary (one client, one time period, one method), that boundary is stated alongside the claim.
Review cadence
Different content types have different decay rates. This site applies the following schedule:
Review schedule by content type
| Content type | Review frequency | Trigger for immediate update |
|---|---|---|
| Evergreen pillar pages | At least every 6 months | A foundational claim changes or is disputed |
| AI-tool and crawler guidance | Quarterly | A major platform changes its announced policy |
| Research pages | When new primary data is published | The cited source revises or retracts |
| Event and role pages | Immediately after a material change | Role ends, event passes, or date changes |
| Utility and legal pages | At least every 6 months | A legal, regulatory, or platform change affects the content |
The “Last reviewed” date on a page is set only when a real review occurred. It is not set automatically on publish.
Correction process
Corrections are taken seriously. If a published claim is inaccurate, the correction procedure is:
Correction steps
Report the issue
Send the page URL, the specific claim, and the evidence for the correction to the contact page at /contact/. Include a subject line starting with “Correction request.”
Review within 5 business days
Every correction request is reviewed against the claim register and the primary source. If the claim is wrong, it is corrected. If the claim is correct and supported, the response explains why.
Update visible on the page
Corrections to substantive claims are noted inline rather than silently replaced, with a brief “Corrected [date]: original language was X; correction is Y.” The reviewed date on the page is updated.
Claim register updated
The claim register entry is updated to reflect the correction, the evidence that prompted it, and the date.
Principle
A correction silently replacing wrong text is less honest than a correction that shows what changed. The visible record of what this site said and when it was corrected is part of the trust structure.
Commercial relationships and editorial independence
Commercial delivery (consulting, speaking, and workshops) runs through AI Marketing Box. This site refers to AI Marketing Box when describing where engagements are scoped and priced. That relationship is disclosed, not hidden.
No content on this site is paid coverage of a third party. No sponsored articles, paid placement, or affiliate-linked claims are published. If that policy changes, it will be disclosed prominently and immediately.
Scope and contact
This policy applies to martinzialcita.com. It does not govern AI Marketing Box, which operates separately. AI Marketing Box has its own data and editorial practices.
Questions, correction requests, and sourcing challenges can be submitted through the contact page at /contact/.