Skip to content

How AI Vanguard reports

Single-author analysis, with sources, inspectable lab evidence and clear limits.

Ehab Al Dissi is responsible for AI Vanguard’s editorial decisions. AI tools assist with research, drafting, code and editing. There is no standing panel of independent expert reviewers. If another person reviews a specific investigation, that article should name them and explain their contribution.

Evidence before certainty

  • Original lab: publish the setup, commands, observed output and limitations. A deliberately constructed failure is an illustration of a mechanism, not a discovered production incident.
  • Reported claim: link the original document and attribute its numbers. Reading a vendor benchmark is different from reproducing it.
  • Worked example: label invented inputs and calculations. A sample pilot plan is not a client case study.
  • Judgment: explain the reasoning and the conditions under which a recommendation may fail.

Sources and dates

We prefer official documentation, source code, advisories, research papers and regulatory records for the claims they can support. Primary sources have interests and limits too. Links let readers inspect the basis of the analysis.

An article’s publication date records when it first appeared. Update notes describe substantive changes. A design change or a simpler headline does not mean every price, model or package version has been re-tested. Dated comparisons should be read as dated snapshots.

Independence and relationships

Coverage should not be bought. Sponsorship, affiliate links, vendor involvement or relevant business relationships must be disclosed where they affect a reader’s judgment. Ehab is involved in Aserva, whose assistant runs on this site. Its presence is not independent evidence for an editorial recommendation.

Corrections

Send the article link, the disputed claim and your supporting source. We investigate errors and add dated notes for material corrections. Routine spelling and layout fixes may be made without a separate note.

Useful without pretending to be universal

Our scripts and checklists have a stated scope. Test them in your environment and review the permissions they need. A health check, an illustrative budget and a laboratory traffic sample each answer a limited question; none proves an entire system is safe.

The Playbook offers practical decision rules. We do not currently publish a validated “Vanguard Benchmark” or a scientifically validated AI maturity score.

About the author · Browse the evidence · Research updates by email

Free operating manual
Get the AI transformation playbook behind this site.

135 pages: frameworks, use cases, governance, ROI, and a 90-day execution plan.

Unlock the playbook →