This page is the editorial policy document explaining how DijitalPi uses artificial intelligence in content production, which checks human editors apply, and what our publishing standards are. DijitalPi is an Istanbul-based, new-generation digital marketing agency that combines the speed of artificial intelligence with human expertise across strategy, content, advertising and data analysis. Our reason for publishing this policy is simple: our readers and clients have the right to know how the content they consume is produced.
Summary: At DijitalPi, artificial intelligence is an assistant that adds speed in research, drafting and data analysis. No content is published without human editorial approval. Official platform documentation takes priority among sources, unverifiable claims do not go live, and additional regulatory checks are applied on sensitive topics such as healthcare.
The fundamental rule of our entire content operation is this:
Artificial intelligence is an assistant that speeds up the content production process. Final content is not published without being checked by human editors for accuracy, originality, brand alignment and user benefit.
At Which Stages of Content Production Do We Use Artificial Intelligence?
We use artificial intelligence at three main stages: research (keyword data, search intent and competitor content analysis), drafting (staged writing tied to an approved brief) and data analysis (processing search and performance data). Topic selection, strategy, source verification, final quality control and the publishing decision always belong to people.
The process works with a division of labour in which roles are clearly separated at every step:
| Stage | AI's role | The human role |
|---|---|---|
| Research and topic analysis | Keyword data, search intent classification, competitor content scanning | Strategy decision, topic selection, approval of the content brief |
| Drafting | Writing the first draft section by section, tied to the approved brief | Direction on structure, language and brand positioning |
| Data analysis | Processing search and performance data, detecting patterns | Interpreting the data and deciding on action |
| Source verification | Listing candidate sources | Confirming every claim one by one against a primary source |
| Publication | No publishing authority | Final approval and publication |
| Updating | Detecting outdated information and unreachable sources | Re-evaluating and revising the content |
This table applies not only to our own site but to all content we produce on behalf of our clients.
How Are Research and Source Verification Carried Out?
Every piece of content is verified against our three-tier source hierarchy before going live. First priority is the official documentation of platforms such as Google, Meta, OpenAI and Anthropic; second priority is peer-reviewed academic research; third priority is credible industry research with an open methodology. Unverified blog claims are not used in our content, however widely they are repeated.
Every source proposed by artificial intelligence at the research stage is opened and checked one by one by an editor: does the source genuinely exist, does it genuinely support the claim, is it current? Sources that cannot be reached or whose content does not match the claim are removed from the draft. Statistics are always given with their context and date; we do not use figures without context.
What Do Human Editors Check?
Human editors check every piece of content on four dimensions: accuracy (every claim, statistic, date and name is confirmed against its source), originality (repetition and template language are stripped out, original analysis is added), brand consistency (language, tone and positioning match brand standards) and user benefit (does the content genuinely answer the reader’s question?).
Ultimate responsibility for editorial processes lies with the team led by our founder Selim Çitil, who has over 20 years of marketing experience. Author information, publication date and update date are clearly stated on every published piece; no unchecked draft is signed with that information.
How Is the Risk of Misinformation (Hallucination) Reduced?
We reduce hallucination risk through four mechanisms: staged production (section-by-section writing tied to an approved brief, rather than long text in a single pass), claim-level verification (every specific piece of information is confirmed against an independent source), separation of producer and reviewer (the system producing the content cannot approve its own output) and removal of the unverifiable (a doubtful claim that cannot be corrected does not go live).
The practical consequence of this approach is that if we cannot verify a piece of information, it comes out of the text however impressive it looks. "It is probably true" is not a publishing standard for us. Fabricated statistics, fabricated customer reviews and unevidenced success claims are strictly forbidden, both in our own content and in client content.
How Often Is Content Updated?
We review our content at planned intervals according to topic type: fast-moving artificial intelligence and platform topics monthly, advertising platform content every two to three months, core guides and pillar content every three to six months, and service pages every six months. Updating is not merely changing the date but re-evaluating the content.
| Content type | Review frequency |
|---|---|
| AI and platform updates | Monthly |
| Google Ads and Meta advertising content | Every 2–3 months |
| Comparison and list content | Every 3 months |
| Pillar content | Every 3–6 months |
| Service pages | Every 6 months |
| Core concept content | Every 6–12 months |
This policy page is subject to the same discipline; the next scheduled review: January 2027.
Is Anything Published Fully Automatically?
No. At DijitalPi no content is published without human approval. Our artificial intelligence systems have no publishing authority; production and publication are two separate steps, technically separated as well. Even if a piece of content passes every stage of editorial control, the final publishing decision is always made by a person.
This rule does not change on projects where we produce high volumes of content. Automation’s job is to prepare the draft and speed up the review process — not to press the "publish" button.
What Additional Checks Apply to Sensitive Topics Such as Health, Law and Finance?
On topics that directly affect a user’s life (YMYL) such as health, law and finance, we apply additional layers of control on top of the standard process: a regulatory compliance check, support of claims solely by primary sources, a ban on wording containing guarantees or promises of certain outcomes, and display of reviewer information alongside the update date.
In healthcare content, for example, compliance with the Ministry of Health's promotion and information regulations is screened as a separate control step: price information, superiority claims, guarantee language and patient testimonials are not used. Medical claims are grounded in peer-reviewed literature; where necessary, an opinion is sought from a specialist in the relevant field. The same principle applies to legal and financial content: such content is for information purposes, does not replace professional advice, and that limit is stated explicitly in the content.
How Are Commercial Relationships and Client Content Labelled?
We do not conceal our commercial relationships: content we produce on behalf of clients is published with the client’s approval on the client’s channels; sponsored and collaborative content is clearly labelled. If DijitalPi appears in a comparison or list article on our own site, the commercial nature of that is not hidden from the reader and the evaluation criteria are presented openly in the content.
In case studies, a client’s name is used only with written permission; without permission, the case is published anonymously with sector information given. In comparison content we do not give unverified information about competitors and do not automatically declare DijitalPi the winner — we explain which option suits whom, according to the need.
Frequently Asked Questions
Is your content written entirely by artificial intelligence? No. Artificial intelligence prepares the first draft in line with an approved brief; topic selection, source verification, editorial control and the publishing decision belong to people. The published text is the version that has passed a human editor’s checks on accuracy, originality, brand alignment and benefit.
Why do you state openly that you use AI? Because transparency is a precondition of trust. An agency that hides its use of artificial intelligence cannot have its claim of "human-supervised production" tested. We explain the whole process, so that reader and client alike can decide knowing what was produced and how.
What should I do if I find an error in a piece of content? Tell us: info@dijitalpi.com. Every reported error is examined by the editorial team; confirmed errors are corrected and the content’s update date is refreshed.
Does this policy also apply to client projects? Yes. The source hierarchy, human editorial approval, hallucination controls and the additional YMYL layers apply to all content DijitalPi produces — the same standard is applied on our own site and on our clients’ sites.
Get to Know DijitalPi
If you are curious how our human-supervised AI content model would work for your brand, let us walk you through the process with examples. Tell us about your project and let us assess it together.
Cookie preferences