Siri AI beats Gemini at the thing that matters most

Now that iPhones have a functional version of their own voice bot with Siri AI, it was natural for me to pit it against Gemini out of sheer curiosity. Since Google had a solid head start, I expected this comparison to be one-sided, with Siri continuing to be the compromised AI bot being the new kid on the block.

But to my surprise, the new Siri AI has fit into my everyday routine more naturally in a short span of time, while Gemini still feels disconnected from the rest of the apps on my phone.

Gemini’s strengths and weaknesses are the same

I’ve talked at length about how incredible Gemini is at handling tasks within the Google ecosystem. In the last decade of covering tech, this is the first time I’ve seen such a tight integration with so many Google services, and I’m all for it. I can ask Spark to complete multi-step research and add the findings to NotebookLM (now called Gemini Notebook) or ask Gemini to make changes inside Sheets without having to learn a hundred obscure formulae. And it rarely disappoints me.

Where Gemini leaves me wanting more is in third-party integrations. Google’s AI is perfect if you live exclusively in Google’s world, but the cracks start to show when you try to use it with a third-party tool you use every day. For instance, I asked Gemini to add a task to TickTick, an app I’ve used for years to manage my day, but it simply refused to play nice. Without confirming with me first, it defaulted to Google Tasks, which I find too limiting to switch to.

It matters where you live

I can bet that your experience with Gemini could be the opposite of mine, and that’s because cross-app integrations are heavily region-dependent, much like Google’s staggered software release schedules.

In North America, Gemini works with services like Instacart to purchase groceries, Angi to hire home service professionals, and Zillow to provide rental home tours. Not being in the region, these services aren’t a part of my daily routine — heck, I only heard about them while researching for this story. Except Canva, not a single third-party app on Gemini’s list is something I use every day.

That is not to say Google is entirely unbothered about India. It counts the market as one of its biggest, with close to a billion active Android devices. That’s why Gemini has been localized to speak tons of native languages, and Google even bundles the AI Pro subscription with certain carrier plans. Despite the effort to make AI widely accessible here, Gemini feels hamstrung by its inability to talk to anyone outside the home.

Siri AI is apparently friendlier

While using Siri AI on an iPhone alongside Gemini on my Pixel, I noticed two distinct usage patterns on both phones. On the Pixel, I intentionally use Gemini to talk to it and retrieve information, and occasionally to get something done (again, within the Google ecosystem). It can read the phone’s screen and collate that data with Personal Intelligence to show results, and, if needed, take action inside a Google app, say, reserving a slot in Calendar. It’s just a more natural-sounding version of Google Search and Assistant, which could already do a lot of this, including sending messages on WhatsApp.

On the other hand, Siri AI already feels more ready to take actions rather than just pull up information. With TickTick, Siri AI can not only run me through the day’s tasks but also add and manage those entries without having to mention the app’s name with each request.

This frictionless operation is why it feels more natural to fire it up than to open my reminders app. It already works with a bunch of local apps I use daily, such as Blinkit (the Indian Instacart equivalent), where it helps me add items to the cart with just my voice — something I use several times a week.

The developers are only half the story

It’s rarely black-and-white how features are introduced to a wider global audience. Google enabling MCP connections won’t automatically push developers to implement every new feature the company announces. It depends on how worthwhile it is for developers and on Google’s intentions regarding specific markets.

We’ve historically seen Google being more laid-back with guideline enforcement — remember how long it took for third-party apps to implement tinted app icons back in the day? Apple is not only stricter in this area but also offers passive incentives, such as App Store pushes for apps that meet its requirements. Additionally, Apple India’s team worked closely with local developers, including those at one of India’s largest food delivery services, to implement Siri AI integration across its offerings. It shows both intention and commitment, something Google needs to desperately learn.

And it does seem like Google is catching up. The company is reportedly planning to make Gemini work with an Amazon-scale Indian online retailer. It does give me hope, though I think it’s time for Google to floor the gas pedal because it’s not just Apple it’s competing against.

How can Google step up?

AI’s conversational nature is no longer the standout feature — it’s not 2024 anymore. What will really make AI intelligent going into 2027 is its ability to take actions within other apps and how seamlessly it does so. With Gemini, that kind of integration is non-existent for me, and Siri AI is filling that gap. And if you’re creating AI workflows more seriously, Claude and ChatGPT are miles ahead of both Apple and Google in terms of wider compatibility and fewer regional restrictions.

Google recently flicked on MCP connections for its Home app, letting you control your smart home devices, say, with ChatGPT, Claude, or any AI that supports the universal protocol. It’s a solid move, but setting up an MCP connection requires a bit of enthusiast-level tinkering, making it a little less consumer-friendly, and a Home Premium subscription, which keeps it from being widely available. Google pulled a classic one-step-forward-and-two-steps-back.

What Google really needs is to bring onboard as many mainstream apps as possible in each market to improve Gemini’s usability for a variety of everyday needs. But more importantly, it must make cross-app interactions with Gemini stupidly effortless for end users. The process should be as seamless for third-party tools as it is for getting stuff done inside Gmail.

Getting rid of that friction would make me want to use Gemini more out of habit. Make me want to use AI more because it’s the more efficient way to do things, rather than expecting me to get addicted to it one random morning.