Apple Sends Nearly 200 Siri Programmers to AI Programming Training, Two Months Before WWDC
Apple plans to send nearly 200 Siri programmers to participate in a several-week AI programming training course to learn how to use AI tools like Claude Code for coding. With only two months left until the new version of Siri is expected to debut at WWDC in June, the core development team will retain about 60 members after training, while another 60 will be transferred to an evaluation group responsible for testing command processing and compliance.
The Siri team has long been known internally at Apple as "laggards," with a bloated workforce and internal political divisions over the past 15 years, and its competitiveness has further declined with the rise of large language models. Other departments, such as Apple’s software engineering, have widely adopted AI programming tools and allocated budgets, while the Siri team has failed to keep pace. Apple has previously restructured the Siri team multiple times, separating it from former AI head John Giannandrea and placing it under the jurisdiction of software chief Craig Federighi, with Vision Pro head Mike Rockwell directly responsible for the product. The new version of Siri will be powered by Google Gemini, supporting more natural conversations, direct answers, emotional support, and task execution, with Apple discussing server hosting operations with Google.
Source: Public Information
ABAB AI Insight
This training initiative exposes Apple's execution inertia constraints in the wave of generative AI. The Siri team has long relied on traditional software engineering paths, with accumulated personnel scale and internal coordination costs forming a systemic burden, leading to its lag in adopting AI tools compared to other departments. The concentrated catch-up before WWDC is essentially an external capability injection to compress the development cycle, while team diversion achieves streamlining, allowing the core group to focus on advancement and the evaluation group to strengthen quality and compliance checkpoints. This restructuring reflects a forced adjustment from a bloated structure to focused execution.
From a technological substitution perspective, the new version of Siri's shift to the Google Gemini model and hosted servers marks a temporary concession of Apple's self-research priority strategy at the voice intelligence layer. A company known for its closed-loop chips and operating systems is outsourcing part of the reasoning and computation of its core user interaction interface to a competitor, highlighting the strength gap between the model layer and infrastructure layer. This dependency reduces short-term development risks but partially shifts long-term pricing power and data control, accelerating the clarity of the division of labor among hardware, software, and models in the AI stack.
Historically, the multiple delays and leadership changes of Siri (with John Giannandrea's advisory period ending soon) are common aging paths for technology financial companies: early advantages transform into organizational inertia, requiring higher restructuring costs when facing paradigm shifts. Through external AI tool training and cross-company collaboration, Apple attempts to catch up in the lagging productivity curve, while wealth distribution will tilt towards participants mastering efficient model integration and review capabilities, with the residual value of traditional internal teams facing reassessment.
Overall, this event points to the profound transformation of organizational forms of large tech companies due to generative AI: when external models and tools can quickly absorb coding labor, internal scale advantages turn into burdens, and the competition shifts towards managing infrastructure dependencies and solving review bottlenecks efficiently, which will continue to reshape the concentration patterns of capital, talent, and power within the industry.