Apple is sending 200 Siri engineers to an AI coding bootcamp. That tells you everything.

Apple is sending nearly 200 Siri engineers to a multi-week coding bootcamp focused on AI tools, according to The Information. The program will teach them to build software using AI coding assistants, tools that the rest of Apple’s engineering organization has apparently already adopted.

Platformer’s Casey Newton put it bluntly: “It’s 2026 and we’ve decided the Siri team should learn how to code.” That’s uncharitable but not wrong. The Siri division has a reputation, per The Information’s reporting, as “a laggard inside Apple” when it comes to AI tool adoption. Other teams at Apple already use AI coding assistants with large budgets. The Siri team didn’t keep pace.

What happened to Siri

The backstory matters for understanding why this bootcamp exists. Apple promised an AI-powered Siri overhaul as part of Apple Intelligence at WWDC 2024. It didn’t ship. In March 2025, Apple delayed the features indefinitely after Craig Federighi and other executives found that “features didn’t work properly, or as advertised, in their personal testing.”

Engineers had been racing to fix a succession of bugs, and some within Apple believed the work might need to be scrapped entirely and rebuilt from scratch. The core problem was architectural: two separate backend systems (one legacy, one advanced) instead of a unified system. Siri’s original 2011 foundation relied on rigid “if-this-then-that” scripts that couldn’t handle the kind of nuanced conversation and task chaining that users expect in 2026.

The leadership fallout was significant. John Giannandrea, Apple’s former AI chief, lost control of the Siri team in March 2025 after Tim Cook reportedly lost confidence in his ability to execute on product development. Giannandrea stepped down in December 2025 and officially left Apple on April 15, 2026, the same day The Information’s bootcamp report dropped. He plans to join startup boards and do advisory work.

Craig Federighi now oversees AI development directly. Mike Rockwell (formerly the Vision Pro lead) runs the Siri team. And in December 2025, Apple hired Amar Subramanya as VP of AI. His resume is telling: 16 years at Google (rising to VP of Engineering and head of engineering for Gemini), then five months at Microsoft as corporate VP of AI. Apple brought in someone who built the competitor’s product.

The Google Gemini deal

The bootcamp doesn’t exist in isolation. In January 2026, Apple signed a multi-year deal with Google for Gemini to power the rebuilt Siri and Apple Intelligence. The deal reportedly costs Apple around $1 billion per year and involves a custom 1.2 trillion parameter Gemini model, eight times larger than Apple’s existing 150 billion parameter cloud models.

The model uses a mixture-of-experts architecture optimized for summarization, planning, and natural language understanding. Critically, the Gemini model weights run on Apple’s own infrastructure, not Google’s servers. No user data is shared with Google, and data isn’t stored after processing. First Gemini-powered features are expected with iOS 26.4 later this spring.

So Apple has the model. It has the infrastructure deal. What it apparently lacked was an engineering team that knew how to build with modern AI tools. That’s what the bootcamp is fixing.

The combination of a trillion-parameter model and engineers who don’t yet know how to use AI coding tools creates an obvious bottleneck. You can’t productively integrate a model you don’t understand into code you’re writing the old way. The bootcamp is the bridge between Apple’s infrastructure investments and the engineering output those investments are supposed to enable.

What the bootcamp actually teaches

According to The Information, fewer than 200 engineers from the Siri team will attend. While they’re in training, roughly 60 engineers will continue active Siri development and another 60 will work on testing and evaluation.

The focus is practical: using AI coding assistants like Anthropic’s Claude Code and OpenAI’s Codex to ship more code faster. These tools, as The Information notes, “have upended the programming profession, allowing experienced software developers to produce far more code than they have in the past.” Some Apple teams outside Siri have reportedly received Claude access with $300/day token budgets, though that claim comes from an unverified social media post.

The timing is everything. WWDC 2026 is scheduled for June 8. Apple is widely expected to announce the long-delayed AI-powered Siri as part of iOS 27, with a general release in September 2026. That gives the bootcamp graduates roughly five months to apply what they’ve learned before the product ships to a billion devices.

The scale of retraining is notable too. Nearly 200 engineers is a significant chunk of the Siri organization, which has “hundreds” of total staff. Apple isn’t sending a pilot group of 20 to test the waters. It’s pulling a third or more of its Siri workforce off active development simultaneously, betting that the short-term productivity loss is worth the velocity gain once they return.

For context, the broader tech industry has been through this transition already. GitHub reported that Copilot users write code 55% faster. Anthropic and OpenAI both report that their enterprise customers see 2-3x throughput gains from AI coding tools. Apple’s Siri team was apparently the last major engineering group at the company to adopt these tools, and the bootcamp is the company admitting that delay cost them.

What this means for you

If you’re an Apple user waiting for a Siri that can actually compete with Google Assistant or ChatGPT, the honest read is mixed. On one hand, Apple has hired the right people (Subramanya from Google/Microsoft), secured the right model (Gemini 1.2T), and is now urgently retraining the team. On the other hand, sending your core engineers to bootcamp two months before WWDC suggests the gap between where Apple is and where it needs to be is larger than the public timeline implies.

If you’re a developer who ships apps on Apple’s platform, pay attention to the App Intents API. That’s the surface where a smarter Siri would interact with third-party apps, and Apple’s historical weakness there (fragile integrations, limited natural language support) is exactly what the Gemini partnership is supposed to fix. A Siri that can chain multi-step tasks across apps would change how users discover and interact with your features. If Apple nails this, App Intents becomes the most important integration surface since widgets. WWDC should tell you whether they’ve actually done it.

Apple’s privacy-first approach to AI (on-device processing, Private Cloud Compute, no data sharing with Google under the Gemini deal) remains its key differentiator. But privacy architecture is a feature, not a product. Users don’t choose Siri because their data stays on-device; they choose it because it works. Or, increasingly, they don’t choose it at all.

2026 is Apple’s make-or-break year for AI credibility. The bootcamp is an admission that they know it.