Apple Siri ‘Campos’: The Gemini Upgrade That Finally Fixes Siri

~20 min read

For a decade, asking Siri a complex question resulted in one infuriatingly consistent answer: “Here is what I found on the web.” That era of digital incompetence ends today.

If you are an iPhone user, you have lived through the “Siri Stagnation.” It has been a decade of frustration. While competitors like ChatGPT and Claude were passing the Bar Exam, writing Python code, and planning complex European vacations, Siri struggled to set two timers simultaneously or understand a sentence if you changed one word. Apple’s strict adherence to on-device privacy, while noble, often came at the steep cost of capability. Siri was a safe vault, but it was an empty one.

Enter Project Campos.

Released in the latest iOS update, “Campos” is the internal codename for Apple’s historic, albeit reluctant, partnership with Google. It integrates Gemini Advanced directly into the Siri interface, creating a hybrid AI architecture that finally delivers on the promise of a true digital assistant. It combines Apple’s local “Personal Context” (what’s on your phone) with Google’s massive “World Knowledge” (what’s on the web).

In this comprehensive guide, we’ll break down exactly what the Siri ‘Campos’ update is, the technical nuance of the privacy handoff, and how to use it to turn your iPhone into a genuine productivity engine.

What is Siri ‘Campos’?

‘Campos’ is the orchestration layer that sits between iOS and Google Gemini.

It is crucial to understand that Campos is not just a “chatbot app” you download from the App Store. It is a fundamental architectural rewrite of how Siri processes human language.

Previously, Siri was a monolithic system that tried (and often failed) to handle every request itself. Now, Siri acts as a Semantic Router. It analyzes the intent of your query before deciding who should answer it.

Local Tier (Apple Intelligence):
- The Domain: Personal data, device settings, and simple logic.
- The Mechanism: If you ask “Play my workout playlist,” “Find photos of my cat from 2022,” or “Turn on Dark Mode,” Siri handles it entirely on-device using Apple’s own small language models (SLMs).
- The Benefit: It is instantaneous, offline-capable, and privately processed on the Neural Engine of your A-series chip. No data leaves your pocket.
Cloud Tier (Gemini via Campos):
- The Domain: World knowledge, complex reasoning, creative writing, and coding.
- The Mechanism: If you ask “Plan a 3-day itinerary for Tokyo based on these four emails and verify the train times against the public schedule,” the local Siri model realizes it is out of its depth. It packages the query (and necessary context) and hands it off to Google Gemini in the cloud.
- The Benefit: You get the reasoning power of a massive data center model without having to open a separate Google app. The answer returns natively inside the Siri UI, indistinguishable from a local response.

The Core Promise: You get Apple’s privacy for personal data and Google’s IQ for world data, without the friction of app switching.

Why it matters: The End of the “Walled Garden” AI

This update matters because it represents a massive strategic pivot for Apple—a company famous for controlling every widget in its ecosystem.

Admitting Defeat (The Pragmatic Pivot): Apple tacitly admitted that they couldn’t build a “World Model” (LLM) as good as Google or OpenAI quickly enough to compete. Partnering was the only way to save Siri from irrelevance. It signals a shift from “We build everything” to “We curate the best services.”
The “App Stitching” Revolution: Campos doesn’t just chat; it uses a new framework called App Intents. Traditional LLMs give you text. Campos can take information from a Gemini answer and perform actions in Apple apps. It can take a generated itinerary text and actually book the calendar slots or open the specific Maps location. It bridges the gap between “Thinking” and “Doing.”
Mainstreaming AI: For 1 billion iPhone users, “using AI” is no longer about signing up for a specialized service or learning how to prompt. It’s just… using their phone. By baking Gemini into the OS, Apple has instantly made advanced AI accessible to everyone from teenagers to grandparents.

What’s New: Key Features

The Campos update brings three specific superpowers to the iPhone that change the daily user experience:

1. Screen Awareness (Visual Context)

Siri can finally “see” your screen. This is a multimodal capabilities leap that transforms how you interact with apps.

Scenario: You are scrolling Instagram and see a photo of a weird, fractal-looking vegetable.
Action: You trigger Siri (long press power button) and simply say, “What is this and how do I cook it?”
Result: You don’t need to screenshot, crop, and upload. Campos takes a temporary snapshot of the screen state, sends it to Gemini Vision, identifies it as a “Romanesco Broccoli,” and pulls up a recipe card. Crucially, it does this as an overlay, so you never leave the Instagram app.

2. The Privacy Handoff (The “Blue Glow”)

Apple solved the “privacy nightmare” of sending data to Google with a distinct UI cue designed to build trust.

The Apple State: When Siri handles a request locally (e.g., setting an alarm), the screen edge glows the standard, multi-color Apple Intelligence gradient.
The Google State: When Siri determines it needs to send data to the cloud (Campos), the glow morphs into a distinct Pulsing Blue, and a subtle haptic tap prompts you.
The Choice: For the first few times, a banner asks: “Do you want to use Google Gemini to answer this?” You can set this to “Always Allow,” but the visual cue remains. You always know exactly when data is leaving the sanctity of your device.

3. Dynamic Island Memory

Siri doesn’t disappear into the ether anymore. If you ask a complex question, the answer minimizes into the Dynamic Island at the top of the screen.

The Workflow: You ask Siri to “Draft a generic email to my landlord about the leak.” You can then minimize that answer, open your Mail app, and drag the text from the Dynamic Island directly into the compose window.
Persistence: It turns Siri into a persistent clipboard manager and research companion that “lives” at the top of your phone while you work in other apps.

Real-World Use Cases

Here is how the hybrid model changes daily workflows, moving from simple commands to complex agents.

1. The “Travel Agent” Workflow

The Ask: “I have a flight to London on Tuesday. Find a hotel near the arrival airport under $200 that has a gym, and email the options to my partner.”
The Process:
1. Apple Layer: Siri searches your local Apple Mail database to find the specific flight number and arrival time (e.g., landing at Heathrow at 4 PM).
2. Handoff: Siri passes “Heathrow Airport” and the time context to Gemini.
3. Campos Layer: Gemini searches Google Hotels for real-time pricing, filtering for “Gym” and “Under $200” near Heathrow.
4. Apple Layer: Siri takes the list of hotels returned by Gemini and opens the Mail compose window, pre-filling the subject line and body.
The Result: A complex multi-step task involving private data (flight) and public data (hotels) finished in 15 seconds.

2. The “Study Buddy” (Visual Math)

The Ask: A student points their iPhone camera at a handwritten calculus problem in a textbook that they don’t understand.
The Process:
1. Visual Analysis: Siri (via Gemini Vision) recognizes the handwriting and converts the image to mathematical notation.
2. Reasoning: It solves the problem step-by-step.
3. Display: Instead of just giving the answer, it displays the logic overlay directly on the camera view using AR, showing how to solve it.

3. The “Coding Companion” (iPad Pro)

The Ask: “Write a Python script to scrape this website for all PDF links and save them to a folder in my Files app.”
The Process:
1. Campos Layer: Gemini generates the actual Python code logic, using libraries like BeautifulSoup.
2. Apple Layer: Siri Campos takes that code block and, instead of just showing text, offers to save it as a .py file directly in your iCloud Drive or open it in Swift Playgrounds.

Step-by-Step: Enabling Campos

This feature is rolling out now, but it requires specific settings to activate the full hybrid capability.

Step 1: Update iOS

Ensure you are on iOS 18.4 or later. Earlier versions of iOS 18 include Apple Intelligence but lack the third-party “Advanced Intelligence Extensions” required for Campos.

Step 2: The Settings Menu

Go to Settings > Apple Intelligence > External Models.

Provider Selection: Select “Google Gemini” as your provider. (OpenAI is also an option, but Campos is optimized for Google Maps/Search integration).
Smart Handoff: Toggle “Smart Handoff” to ON. If this is off, Siri will never ask to use the cloud, severely limiting its IQ.

Step 3: Account Linking

Note: If you have a Google One AI Premium subscription, Campos gets access to Gemini Ultra (the smartest, reasoning-heavy model). Free users get Gemini Flash (faster, lower latency, but less capable at complex logic).

Limitations (The “Hybrid” Friction)

It’s not seamless magic yet. There are friction points inherent in stitching two giant tech ecosystems together:

The “Handoff Lag”: When Siri decides to switch to Gemini, there is a noticeable 1-2 second pause while the connection handshakes and the data uploads. It is not as instant as a local command.
Privacy Gaps: If you use Gemini, you are subject to Google’s data policy for that specific interaction. Apple strips your IP address before sending the request to mask your identity, but the content of your query (e.g., the photo you sent) goes to Google servers for processing.
No “Deep” Personal Data: Gemini doesn’t know your Health data, iMessage history, or Photos library. Apple refuses to send that deep context to the cloud for security reasons. Consequently, Gemini can’t say “Plan a workout based on my heart rate yesterday” because it is blind to the HealthKit data.

Conclusion

Siri ‘Campos’ is the update that saves the iPhone from obsolescence.

By swallowing their pride and partnering with Google, Apple has given users the best of both worlds: a phone that knows you (Apple) and a phone that knows everything else (Google). It transforms Siri from a glorified egg timer into a legitimate, multi-modal research assistant.

The “Walled Garden” has a gate now, and it opens to the rest of the world’s intelligence.

Ready to try it?

Update your phone, find a complex image, trigger Siri, and ask: “What am I looking at?” The answer might just surprise you.

FAQs

Q: Is Campos free?

A: Yes, the basic integration uses Gemini Flash and is free for all users. Heavy users who want advanced reasoning capabilities can link a paid Google One subscription to access Gemini Ultra.

Q: Does Google see my personal data?

A: Only the specific query you send (e.g., “What is this plant?” and the photo of the plant). They do not see your contacts, photos, location history, or Apple ID details unless you explicitly include them in the text of the prompt.

Q: Can I use ChatGPT instead?

A: Yes. Apple allows you to swap “Campos” (Gemini) for OpenAI’s ChatGPT-4o in settings. However, early tests suggest the system integration with Google Maps, Flights, and YouTube tends to be tighter with Gemini due to the specific API hooks Apple built for Campos.

Q: Does it work on HomePod?

A: Not yet. The “Campos” update is currently restricted to devices with the A17 Pro chip or M-series chips (iPhone 15 Pro/16 and iPads) due to the heavy local processing required for the “Semantic Router” logic that decides when to use the cloud.