Useful guidance needs evidence, context, and clear limits.
Our website and blog help operations and finance leaders think through AI transformation and workflow decisions. These standards describe how Clairvance approaches research, drafting, examples, and corrections.
Published by Clairvance · Standards updated .
Follow a claim back to its source.
We prioritize original research, official product documentation, public filings, government publications, and other sources that directly support a claim. A search excerpt or another article's summary is a starting point for research; the linked source should support the statement itself.
- Give statistics their scope: source, date, geography, population, and definitions when those details affect interpretation.
- Separate what a source reports from our analysis or recommendation. Explain assumptions behind an estimate.
- Check time-sensitive claims such as pricing, product features, rules, and market data against the relevant source when preparing or materially revising a page.
- Avoid treating a vendor's claim, a survey result, or an industry average as a promised outcome for a reader's business.
When sources disagree or do not answer the question, identify the uncertainty. A source link supports the nearby claim; it does not imply that the source endorses Clairvance.
AI can support the work. It cannot supply the evidence.
Clairvance uses AI tools to assist with research, organization, and drafting. AI-generated text is a draft input. Our standard is to check factual statements against sources, examine the reasoning, and remove unsupported claims before publication.
Review includes the page title, descriptions, visible copy, and structured data. Those elements should describe the same content and should not turn a qualified statement into an absolute claim.
Organization-level attribution identifies Clairvance as the publisher. We do not attach a fictional person, credential, or expert-review label to imply human sign-off. A named author or reviewer is identified only when their participation and relevant background can be accurately represented.
If a method, model, or automated process materially shapes an analysis, the page should explain that method and its limits. Generated images, transcripts, calculations, and summaries should never be presented as evidence of an event or test that did not happen.
Keep proposed designs separate from actual results.
An illustrative workflow explains how a system could work. Its inputs, decisions, and outputs may be fictional. These examples are labeled as illustrative and should not be read as client case studies, live demonstrations, or measured performance.
- Actual client work requires evidence for the claimed scope and permission for identifying details before publication.
- Outcome claims need a defined measurement, period, baseline, and relevant limits. Estimates and goals must be labeled as estimates and goals.
- Testimonials, affiliations, certifications, and partnerships must be genuine and authorized. We do not invent them to fill a page.
- When a number is used to demonstrate a calculation, identify it as an assumption rather than a reported business result.
Our service content describes proposed engagement activities. Specific deliverables, systems, responsibilities, price, and timing belong in the agreed project scope.
Make recommendations specific enough to assess.
A useful recommendation explains the workflow it fits, the information it requires, the exceptions it may encounter, and the decisions people still own. It should consider whether process changes, existing software, or conventional automation can address the problem before recommending a custom AI system.
For new articles and substantive updates, we compare the complete answer with relevant published guidance and product documentation. Each article should add a usable decision method, worked example, or other original contribution. We review the full body against its headline, check material claims and calculations, and record the actual review, including automated review. A search ranking or a successful software check is not proof of factual accuracy.
Our articles inform operational decisions. They cannot establish your organization's legal, regulatory, financial, or technical requirements without the relevant context and review. Where specialist judgment is material, the content should make that dependency clear.
Clairvance offers consulting and implementation services, so our recommendations have a commercial context. Readers can assess our service offer and the cited evidence separately. We do not promise rankings, AI citations, financial returns, or universal performance from a workflow design.
An updated date should reflect a real change.
A publication date records when an article first appears. An updated date should reflect a substantive revision, such as a corrected claim, changed guidance, or refreshed evidence. We do not change a date solely to make unchanged content appear current.
A material correction should explain what changed in a visible note when that context affects the reader's understanding. Minor spelling and formatting fixes do not require a new publication date. If current evidence cannot support a claim, revise or remove it rather than preserve a confident but outdated answer.
These are our publishing standards, not a certification that every historical page has been fully re-reviewed. A page's own citations, dates, and limitations remain the best place to assess its claims.
Help us identify a problem precisely.
Use our contact form to report a factual error, broken source, unclear attribution, or concern about an example. Include the page URL, the passage involved, and a supporting source or explanation. Please keep private client or personal information out of the message.
We assess the specific claim against the available evidence and revise, clarify, or remove it when a correction is warranted. A report does not automatically establish that the original statement is incorrect.
Send a correction →