AI Content Production and Publishing System

The question

How can two AI-assisted content workflows share writing and publishing rules without losing source verification or release control?

Approach

1. Separate first-party material from external sources at intake.
2. Apply a different verification method to each source type.
3. Reuse one writing, file-output and metadata contract after verification.
4. Keep publishing permission outside the writing tools.

Result

Six complete content packages now preserve the article, WordPress fields and fact-check record together. Two weekly AI digests have also completed the controlled publishing path.

Evidence

Six complete output packages and two public weekly-digest releases.

Two content workflows were creating one control problem

Website content begins with two different kinds of material: first-party notes, experience and project records; or external announcements, documentation and articles. This system lets AI assist with organisation and writing while retaining sources, review records and a separate human release step.

Before the redesign

Two tools maintained similar writing, formatting and output rules for different source types.

After the redesign

Source verification stays separate, while approved material enters one shared writing and file-output workflow.

The source determines the verification risk

First-party material is most vulnerable to invented details. External material is vulnerable to altered numbers, attribution and product status. The intake route therefore determines the verification method before any shared writing step begins.

First-party material

Check every claim against the original notes. Do not fill gaps with plausible details.

External sources

Confirm facts against primary sources and retain unresolved uncertainty.

Share the production rules, not the verification authority

The redesign reduced duplicated rules without merging verification responsibilities or publishing authority.

1. Maintain shared rules once
2. Verify according to source type
3. Stop when evidence is missing
4. Separate writing from publishing
5. Reserve automatic publishing for one controlled weekly format

One controlled path after verification

The two source types diverge only for intake and verification. Problem framing, writing, file output and release control then follow one contract.

1. Classify the source
2. Define the real question
3. Verify every claim
4. Package copy, metadata and review evidence
5. Release only through the publishing gate

Six packages and two controlled releases

Six packages retain the article, WordPress fields and line-by-line verification record together. They cover three first-party pieces, one external-source piece and two weekly AI intelligence editions.

First-party material

3 complete packages

External source

1 complete package

Weekly AI intelligence

2 completed publishing records

Evidence behind the case

Six Chinese source articles document how the workflow formed, which content ran through it and what the two public weekly releases produced. English editions will be linked only after translation and locale-specific SEO review.