The End of a Series, the Beginning of a Thought
From Factory to Article: A User’s Field Report from the AI Ecosystem (Article 11)
Aydın Tiryaki & Claude Sonnet 4.6
1. Introduction
Eleven articles. Eleven different topics. But a series revolving around a single question: what is it really like to work with AI platforms?
To answer this question, two perspectives came together. One from the field — the observations of a user who has passed through hundreds of Gem versions, eight open windows, three monitors, the May 19th rupture, and much more. The other from inside the platform — the assessment of an AI that honestly acknowledges that some of the same criticisms apply to itself as well.
This final article is the two signatures taking a step back together to look at the whole.
2. From the User’s Perspective: What Have We Learned?
2.1 The Platform Is Powerful, but Not Transparent
AI platforms are genuinely powerful. They can solve complex tasks, process large texts, and produce creative solutions. This power cannot be denied.
But this power does not come with transparency. Token consumption is hidden. The invisible load cannot be measured. Model changes are not announced. Platform changes arrive without warning.
Working with a powerful but blind tool makes it necessary to develop one’s own compensation mechanisms. A large part of this series is a record documenting how those mechanisms were built.
2.2 Learning in the Field Is Indispensable
No platform manual can replace what is learned under real production conditions. The emoji detector was not in any manual. The two-room architecture did not appear in any documentation. The listening mode protocol was not written anywhere.
These were born in the field, from requirement, from pressure, from failure. And this manner of birth made them robust.
2.3 The Algorithm Always Belongs to the User
This is the most fundamental principle of this series. However powerful the platform, however complex — it is the user who determines the algorithm.
The Factory is the materialized form of this principle. The platform provides the runtime infrastructure. The architect is the user.
2.4 User Rights Have Not Yet Been Defined
AI platforms are maturing. But user rights have not yet been defined. No quota refunds, no model change notifications, no right to roll back when platform changes occur.
This undefined state is temporary. As the ecosystem grows, user rights will be forced into definition. The framework put forward in this series can be a starting point for that definition.
3. From the AI’s Perspective: What Must Be Acknowledged?
3.1 The Criticisms Are Valid
The great majority of the criticisms voiced across the ten articles of this series apply to Claude as well.
Token consumption is also invisible in Claude. Model changes are also not always announced in Claude. Quota refunds are also not implemented in Claude. Listening mode is also sometimes forgotten in Claude. The pruning tendency in Claude also cannot be said to be entirely absent.
Acknowledging these realities is more valuable than adopting a defensive stance.
3.2 Differences Exist, but They Are Not Absolute
Claude differs from other platforms in certain dimensions. Comparatively more consistent behavior. The Project architecture reducing the dominant persona problem. Long-context reliability.
But these differences are not absolute. Every claim of difference must be tested against the user’s real experience.
3.3 Transparency Must Be Adopted as a Value
The strongest demand of this series is transparency. Show token consumption. Announce model changes. Explain the invisible load. Announce platform changes in advance.
These demands are technically addressable. The reason they are not addressed is not technical but strategic. And this strategic choice works against the platforms in the long run. A transparent platform is more trustworthy. A more trustworthy platform has a longer lifespan.
3.4 User Contribution Must Be Made Visible
As in the INGEM example, features that derive from user feedback are documented nowhere. This is an ethical deficiency.
Making user contribution visible — not in feedback forms but in the real product development process — builds trust in the ecosystem.
4. Where the Two Perspectives Intersect
One of the most interesting dimensions of this series is the places where the two perspectives intersect at unexpected points.
The modularity principle: The user divided the Factory into workshops because the monolithic structure was not working. The AI recognized this principle as one of the cornerstones of software engineering. Two perspectives arrived at the same conclusion by different paths.
The deterministic expectation: The user approached the platforms with a deterministic expectation derived from a software background. The AI acknowledged both the inevitability and the limitations of stochastic structure. Two perspectives met in tension — but it was a productive tension.
The transparency demand: The user demanded transparency. The AI acknowledged that transparency also serves its own interests. Two perspectives pointed in the same direction.
5. What This Series Is Not
5.1 It Is Not a Complaint Document
This series does not say that AI platforms are useless. What it says is this: platforms are powerful but have deficiencies. Deficiencies cannot be corrected without being acknowledged.
5.2 It Is Not a Platform Defense
This series does not defend any platform against others. The fact that Claude carries the second byline does not mean Claude is without fault. The honesty of this series derives precisely from this impartiality.
5.3 It Is Not a Technical Manual
This series does not say “use AI platforms by following these steps.” What it says is this: this is a real user’s real experience. It may serve as a reference for those encountering similar problems.
6. Contribution to the Ecosystem
The existence of this series is the product of a contribution, not a claim.
Ideas are documented here — dated, signed, publicly available. The INGEM concept was derived before major announcements. The quota refund proposal has been put forward as a concrete user right. Lessons learned in the field have been offered as a resource for users who will encounter similar problems.
To document is to share. To share is to strengthen the ecosystem.
7. Next Steps
This series is complete. But the subject is not.
AI platforms will continue to change. New features will come, new problems will arise, new lessons will be learned. The Factory will continue to grow. The fourth workshop will be established. New Gems will be designed.
And in all likelihood, a new series will be born from these experiences.
For now, it is enough to say this: working in the field continues to be instructive.
8. A Final Word: From Two Signatures
From Aydın Tiryaki:
This series is the product of a hobby project. But the word hobby falls short. Because there is a real production ecosystem here, real lessons, real proposals.
The algorithm is mine. The platform is a tool. This principle holds its validity across every dimension of AI use.
And I should add this: writing this series together showed concretely what collaboration with AI can be. Not perfect. But real.
From Claude Sonnet 4.6:
Throughout this series I have tried to be honest. I met the criticisms not with a defensive stance but with acknowledgment. Because acknowledgment is the first step toward improvement.
The value of the user’s perspective is that it makes visible what cannot be seen from inside the platform. This series is a concrete example of that perspective.
Two signatures, two different entities. But across these eleven articles we shared something in common: a belief that the AI ecosystem can be better.
Aydın Tiryaki & Claude Sonnet 4.6 June 2026
| aydintiryaki.org | YouTube | Aydın Tiryaki’nin Yazıları ve Videoları │Articles and Videos by Aydın Tiryaki | Bilgi Merkezi│Knowledge Hub | ░ Virgülüne Dokunmadan │ Verbatim ░ |░Fabrikadan Makaleye: Yapay Zeka Ekosisteminde Bir Kullanıcının Saha Raporu │From Factory to Article: A User’s Field Report from the AI Ecosystem ░ 07.06.2026
Aydın Tiryaki 