Review of the First Three Sections
Date: 2026-05-24
Scope of review:
- Introduction
- Operating Model
- AI Products Portfolio
Verdict
The first three sections already look like a coherent methodology rather than a scattering of notes.
The logic has become clear:
- Introduction explains the problem and the documentation map.
- Operating Model describes the overall management framework: principles, roles, entities, processes, and rituals.
- AI Products Portfolio reveals the reusable product layer: how to avoid producing one-off solutions for every case.
But before the final polish, we need to go through the terminology, numbering, and individual repetitions.
1. Consistency of style
Score: 7/10
What is already good:
- The terms
AI function,AI initiative,AI product,portfolio,delivery,stage gate,adoptionare generally used consistently. - The roles in
Operating ModelandAI Products Portfolioare already reasonably separated: director, portfolio lead, product owner, initiative owner, etc.
What should be fixed:
- In the glossary the term is called "AI Product" (reversed word order), while in the other sections it is "AI product". This needs to be brought to
AI product. - There are variants
AI Conveyor,AI-conveyor,AI Conveyor Framework. It is better to choose one brand variant: AI Conveyor as the product/mechanism, and not to useAI-conveyor. - There is a mix of English and Russian variants of the term for delivery. The recommendation is outdated: the canonical term for delivery is delivery track.
stage gateandStage-Gateare used in parallel. Recommendation: usestage gatein prose and reserveStage-Gateonly for model names or page titles.
2. Storytelling
Score: 8/10
Right now the sequence is good:
- Introduction answers the question "why do we need the system at all".
- Operating Model answers the question "how the management of AI adoption is organized".
- AI Products Portfolio answers the question "how to avoid producing separate solutions for every case".
This is logical for the reader: first the problem, then the overall framework, then one of the key frameworks—reusable AI products.
What gets in the way:
- In
introduction/ai-conveyor.mdxthere is a sense of an old layer: it is not always clear whether this is AI Conveyor as a methodological mechanism or as an application/product. It is better to explicitly separate the wording: AI Conveyor as an application / operational mechanism. - In
operating-model/entities.mdxthe numbering jumps:4. AI initiative, then7. AI product,8.... This looks like a leftover from a draft. It needs to be renumbered. - In
ai-products/product-model.mdxthere is an editorial phrase like "From the old version of the page, what matters here is...". It cannot be left in a public methodology.
3. Duplication of ideas
Score: 7/10
There is useful duplication:
- The definition of
AI productis repeated inentities,product-model,roles,product-owners,product-lifecycle. This is fine, because pages should read standalone. - The idea "do not produce separate solutions for every case" is repeated often, but it is central to the product section.
There is excessive duplication:
ai-products/roles.mdxandai-products/product-owners.mdxpartially overlap on the AI product owner role. This is acceptable as long as the focus is kept:roles= a map of all roles;product-owners= a deep dive into a single role.
product-model,product-lifecycle, anddelivery-funnelall repeat the platform and applied products. It is better to leave a short reminder in each and keep the main detailed analysis inproduct-model.- The anti-patterns "products live on their own / there is no owner / pilots do not scale" are repeated on almost every page of the product section. This is appropriate, but the wording can be varied a little.
Recommended edits after completing the remaining sections
- Bring the glossary in line with the current terms:
AI product,AI Conveyor,delivery track,stage gate. - Remove the editorial phrase from
ai-products/product-model.mdx. - Renumber
operating-model/entities.mdx. - Align the labels in the sidebar: replace
Product LCwithLifecycle. - Check
introduction/ai-conveyor.mdx: separate "the conveyor as a methodological mechanism" and "the AI Conveyor application".
Final assessment
The foundation is good. This is no longer "a pile of pages", but a clear methodology with three levels:
- why the system is needed;
- how the operating model is organized;
- how to manage reusable AI products.
It is worth returning to this review after completing the remaining major sections, in order to align terminology and remove repetitions across the documentation as a whole.