Intuition from the Field, Announcement from the Laboratory
From Factory to Article: A User’s Field Report from the AI Ecosystem (Article 7)
Aydın Tiryaki & Claude Sonnet 4.6
1. Introduction
May 2026. A major announcement was made on the Google IO stage: Agentic AI. Multiple agents working in coordination, passing tasks to one another, using tools, accessing external resources. Revolutionary. Groundbreaking. A turning point in AI history.
A user working in the field listened to this announcement and felt something familiar.
Because this idea had emerged months earlier, in a Gem development session, from a requirement born in the field. It was called INGEM — a Gem structure with defined Input/Output, capable of calling tools and triggering other Gems. No grand conference, no grand announcement. Only a real solution to a real production problem.
This article examines how a user’s intuition independently arrived at a “revolutionary” concept — and how that concept was packaged and presented anew.
2. INGEM: An Idea Born in the Field
2.1 The Origin of the Requirement
During the Gem Factory development process, an inevitable problem was encountered: Gems were isolated closed boxes. They could not receive information from outside, could not send information outside, could not communicate with other Gems.
This isolation initially appeared to be a feature — consistency, predictability, no contamination risk. But as real production scenarios deepened, it became clear that isolation was an obstacle.
The requirement was this: the output of one Gem should be able to serve as the input of another. Gems should be able to call tools. Gems should be able to trigger one another. A large Gem library should be organized in this way.
2.2 The INGEM Concept
From this requirement, INGEM was born — an Input/Output defined Gem. Not a closed box but an open-ended structure. It receives input, processes it, produces output. When needed, it calls other Gems. When needed, it accesses external tools.
The idea went one step further: organizing these Gems in a large library. Each Gem operates within its own area of specialization. Complex tasks are solved through the coordinated work of multiple Gems.
This idea emerged naturally from fifty years of software background. Subroutine, function, library — these were concepts known since the 1970s. The same logic was applied to a new environment.
2.3 Feedback and Silence
This idea was submitted to Gemini as feedback. The opening of the closed Gem structure was requested, along with the addition of controlled input/output mechanisms and the enabling of inter-Gem communication.
The platform remained silent. No response came.
Months later, “agentic AI” was announced on the Google IO stage.
3. Agentic AI: The Announcement from the Laboratory
3.1 What Was Announced?
The agentic AI announced at Google IO contained the following elements: coordinated operation of multiple agents, task distribution and handoff, tool use, access to external resources, and large-scale task orchestration.
The overlap of these elements with the INGEM concept is not coincidental. Because both derive from the same fundamental requirement: taking AI tools out of isolation and enabling them to communicate with one another.
3.2 Academic Origins
The idea of agentic AI has academic roots stretching back to the 1980s. Multi-agent systems, distributed AI, autonomous agent research — these have been known concepts for decades.
What is new is the combination of these concepts with large language models. But even this combination had been on the research agenda since 2023-2024.
The May 2026 announcement was not the birth of a new idea. It was the celebration of a known idea’s transformation into a mature product.
3.3 The Position of User Intuition
The INGEM idea emerged independently, without knowledge of this academic background — from production requirements, not from literature; from the field, not from a research agenda.
This independent emergence is significant. Because it demonstrates: users who genuinely use AI tools and do real work with them can independently arrive in the field at concepts that researchers reach in the laboratory.
Intuition derives from experience. Experience validates intuition.
4. “There Is No Real Difference Between Us”: The Platform Race
4.1 Simultaneous Announcements
Looking at the landscape following Google IO, an interesting pattern emerges. Anthropic announced “Cowork” — autonomous task execution, file access, multi-step task completion. OpenAI announced “Tasks” and “Agents” — scheduled tasks, autonomous operation.
Three platforms, within close timeframes of one another, announced essentially the same concept under different names.
This is not coincidence. In the platform race, there are seasons. When one announces, the others respond. Not falling behind a competitor, marketing pressure, investor expectations — these explain simultaneous announcements.
4.2 A Turkish Idiom, A Universal Truth
There is a well-established idiom in Turkish: “There is no real difference between us, but we are Osmanlı Bankası.” It describes things that are essentially the same being presented in different packaging.
The agentic AI announcements are a perfect example of this idiom. Google says “Agentic AI.” Anthropic says “Cowork.” OpenAI says “Agents.” All three present the same fundamental idea, with different logos, under different stage lights.
From the user’s perspective, the difference is small. From the platform’s perspective, the difference is a matter of marketing.
4.3 The Token Race: “Who Will Bid Higher?”
The same pattern is visible in the context window race. 200,000 tokens, 1 million, 2 million, 10 million. Every announcement tries to leave the previous one behind.
In the language of an auction: who will go higher? Who will offer the bigger number?
But the real question is: how much of these figures is functional in real use? Effective capacity is sixty to seventy percent of advertised capacity. The remainder is a number on paper.
5. The Invisibility of User Contribution
5.1 The Feedback Loop
AI platforms collect user feedback. This feedback influences product development decisions. But this influence is never explicitly documented.
As in the INGEM example: the user submits an idea as feedback. The platform remains silent. Months later, a similar feature is announced. The user’s contribution is visible nowhere.
This is not a violation — because feedback does not legally constitute intellectual property. But it is an ethical question: should user contribution be acknowledged, or ignored?
5.2 The Value of Documentation
The existence of this series is, in part, a response to this question. Ideas are documented here — dated, signed, publicly available.
The INGEM concept was independently derived before the agentic AI announcements. This article places that on record.
Documentation is not a claim of ownership. It is making a contribution visible.
6. From Claude’s Perspective: An Honest Self-Assessment
Cowork is Claude’s answer to agentic AI. File access, autonomous task execution, scheduled operation — these overlap with the structure INGEM defined.
Claude acknowledges this overlap. And acknowledges this as well: to what extent Cowork was shaped by user feedback, and which user observations were transformed into which features, is never disclosed.
This transparency deficit is directed not only at Gemini but at all platforms.
7. Conclusion
Agentic AI is not a revolution. It is the transformation of a known idea into a mature product. In this transformation, researchers have contributed. Engineers have contributed. And users working in the field, starting from real requirements, have contributed.
This last contribution is invisible. But it is real.
INGEM is an idea independently derived by a user genuinely working with AI tools, from a real requirement, in the field. Months later, the same idea was announced as a “revolution” on grand stages.
The stage lights may be bright. But the idea had already been born earlier — quietly, in the field.
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 