Summary of Key Points
Justin, a Google employee, created a tool called gws that connects AI with Google’s Workspace software. The tool became incredibly popular in the tech community (with 28,000 GitHub stars) but was fired for using the Google brand without following the proper approval process. Ironically, just two days before his dismissal, Google announced the release of an official tool with similar functionality. This incident is not just a conflict between an employee and a company; it also exposes a deeper struggle within large corporations in the AI era: they talk about innovation on the surface but are still adhering to outdated systems. New technologies (such as AI agents) are challenging the control over traditional products, and these companies’ organizational structures are not yet prepared to respond to such changes.
Detailed Explanation
1. Why is this tool so impressive? It opens the door for AI to “work directly”
The gws tool essentially acts as a translator for AI, making it possible to interact with Google Workspace (which is similar to domestic tools like Lark or DingTalk). Normal operations in Workspace involve clicking on web pages and filling out forms, which are suitable for humans but not for AI. gws converts these complex backend interfaces into commands that AI can directly execute, giving it a kind of “universal key.”
- Previously, AI could only offer suggestions: For example, it might help you draft an email, but you still had to click the send button yourself.
- Now, AI can take action on its own: It can read Gmail emails, check calendars, create documents, edit spreadsheets, and manage files—almost as if AI has directly entered your office system to complete tasks for you.
- Even better: automatic updates: When new features are added to Workspace, gws automatically adapts without the need for engineers to develop new versions.
This tool is more than just a simple addition; it creates a “fast lane” between AI and office software, enabling AI to move from being an assistant to an executor.
2. Popular but fired? Compliance is the cover story; control is the real issue
On the surface, Google’s legal department cited the use of the Google brand without approval. However, the real reason lies in the power struggle of the AI era:
- Traditional products control users through their interfaces: Google’s office software interfaces are designed with commercial considerations in mind—e.g., splitting annual fees into monthly payments to make them seem more affordable.
- AI tools bypass these interfaces: gws allows AI to directly interact with the system, eliminating the need for users to see the Google-designed interfaces, which could potentially undermine commercial strategies like paid upgrades or advertisements.
- Companies fear losing control of user access: Justin’s tool suggests that users don’t have to follow Google’s prescribed paths, which touches on a very sensitive issue for large companies: who controls the user interface controls the commercial value.
Therefore, the dismissal was not due to the quality of the tool but because it threatened the existing order.
3. External competitors step in: If you don’t value internal innovation, others will take over the market
After the incident, Peter Stenberg, the creator of the popular OpenClaw tool, mocked Google and released his own similar tool called gog. This highlights a common problem for large companies:
- Internal innovation faces many barriers: Organizational processes, legal approvals, and departmental interests can all prevent innovation from succeeding.
- External innovators are unencumbered: They don’t have to follow company rules and can quickly seize market opportunities.
Justin’s tool was stifled internally, but external competitors stepped in immediately. This is exactly what large companies fear: internal innovation stalls while external rivals move ahead.
4. Google’s loss of key AI talents: Internal strife during the AI transformation
This is not an isolated incident. Many other top AI experts at Google have left recently:
- Norman Szalzer, co-author of the Transformer architecture (rehired by Google for $2.7 billion), joined OpenAI.
- A core scientist from DeepMind (contributor to the Gemini model) also moved to Anthropic.
- Adi Osmani, former head of Google Cloud AI, also left the company.
While the reasons for their departures vary, it’s clear that Google’s internal structure and roadmap are not keeping up with technological advancements. Some want to push forward with innovation, while others prefer to stick to the status quo, leading to a loss of top talent.
5. A common dilemma in the AI era: Should innovation be “controllable” or “disruptive”?
This is a problem that all companies will face:
- Companies want controllable innovation: They allow innovation but only if it follows their plans and does not affect existing business operations.
- True innovation is often uncontrollable: It emerges from user needs, is validated by the market, and then forces companies to recognize its value. By the time they realize its potential, it may have already broken free from their control.
Justin’s tool exemplifies this: it became successful despite not following the official process, forcing Google to either accept it (and adjust its policies) or eliminate it (to protect its interests). Google chose the latter option, but at the cost of losing talented employees and facing external competition.
This dilemma has no simple solution. It highlights that in the AI era, large companies that cannot break free from outdated structures may be overthrown by their own innovators or external competitors.
Final Thoughts
Justin’s story serves as a reminder: AI is not just about adding new features; it requires rethinking the fundamental logic of products. For individuals, this means being cautious of being eliminated due to following rigid procedures. For companies, it means considering how to balance control and innovation. In the AI era, the competition will not be about who can preserve the old order but who can adapt to new rules more quickly.