I waited to see how well (or not) this worked in the real world, before leaning in with my take on a potential crackdown on spam communication. Airtel’s rolled out something that’s essentially first of its kind, this past week (you’d probably have noticed it too), in an effort to curb spam calls and messages destined for users on their network. Full marks for effort, without doubt. You’re thinking, a Truecaller for the entire network? It’s actually more complicated than that…and that’s before I speak with you about Truecaller itself.
Within hours of Airtel announcing this, I started noticing “suspected spam” labelling on certain calls and spam messages (well, a web link that reads indlapost.com is surely a red flag). On Android phones and the Apple iPhone, the call screen has this labelling. For messages, text begins with this labelling.
Airtel’s rolled out something that’s essentially first of its kind, a network wide spam filter
Good start, but that’s where the cracks began to show. Many, and I mean many, random calls came through without any labelling (insurance sales calls, pitches for loans and even SMS inviting me to buy a commercial property; apparently one adjacent to an Apple Store in some mall). And herein lies the problem with Airtel’s initiative, no matter the intent. Therefore it becomes important to build on what they’ve started.
- As a user, I have no input or manual intervention in marking (or correcting a wrong label) any call or message. It’s purely Airtel’s execution. A complete opposite to Truecaller, which gives users complete control for each call—marking it as spam (with further details too) or marking it as not spam.
- While Airtel says there are many factors that decide why and how any call or message is labelled, it’s still largely an opaque process. We don’t know how many ticks on a checklist are needed. Quite why so many calls from ICICI Lombard Insurance filter through these so-called measures, is beyond me.
- Then there’s the matter of perception. I’ll take the example of ICICI’s insurance calls. For me, they are spam, because I have not requested a call-back (or 7 call-backs a day). But for someone who has to buy or renew a policy or needs help with a claim, these will not be spam. This will be true for every single business that relies on some level of calling to existing and potential customers. And that neatly leads me to the next point,
- Third, is the fact that Airtel’s simply “suspected spam” labelling comes with no further details. This will (and it’s already being mentioned on social media) mean users may ignore calls they were otherwise supposed to receive. From your insurance company or car dealer for the insurance renewal premium reminders. From a service center that’s been assigned to fix your refrigerator. From an installer who has to put together the study table you’d ordered on a furniture website. Quite unlike Truecaller, which has a reference (and therefore as a user, you’re better informed to decide whether to pick the call or not).
I could point out that Airtel could have perhaps deployed their Thanks app (it’s available on Android and iPhone) for descriptors. It’d have given them parity with Truecaller in terms of utility. Perhaps still not the massive spammer database that Truecaller has refined over many years. That would have been a smart way to push that app’s downloads (I’m surprised no one thought of that). Maybe that’ll happen in due course.
Once Apple rolls out iOS 18.1 in the coming weeks, Truecaller for iPhone is set for a massive update
Now speaking of Truecaller. It has been the go-to app for anyone who’d wanted call identification on their phones. Worked much better on Android all these years (Samsung, Google and many other phone makers tried integrating third party apps too into their dialler apps; never worked as well), purely because Apple had limited access Truecaller had to their CallKit layer. Yet, the company found a way to still offer some sort of caller ID despite those restrictions. Now things are about to change, significantly.
With iOS 18, specifically iOS 18.1 and the importance of the .1 that I had spoken about last week, Apple’s finally unlocking access to a lot of underlying layers of an incoming call on an iPhone. Truecaller for iPhone is set for a massive update. The specifics of which will become clear once iOS 18.1 and Truecaller’s expected app update rolls out at some point in October. Basically, expect Truecaller to be as powerful at call identification and spam blocking on an iPhone, as it has been on Android over the years.
HEADWAY
If you keep a promise, I’ll do too. Since you are here for our conversation this week, I haven’t forgotten that we must discuss the important stuff from the Meta Connect keynote from a few days ago. My biggest takeaway from the annual conference was none of what was actually announced (surprise, surprise!), but actually what wasn’t. Okay, I’ll meet you mid-way—it finds a link in Meta’s augmented reality (AR) push, that finds a talking point with the Orion AR glasses that had their first moment in the sun. They won’t be going on sale at this time, but it’s a glimpse at what’s possible and what Meta is working on. Orion takes forward the Ray-Ban Meta smart glass legacy, by adding a visual element to what a wearer sees. And that’s my point.
Can Meta build an AR platform, the kind that doesn’t exist even though Snap and Apple have been trying with AR hardware and the developer ecosystem? It’s been a while since Meta (you’d probably remember those pursuits under the Facebook banner) gave the world a platform that mattered to the masses. The metaverse didn’t work. Could AR be that platform, something that would intersect the workplace, our communications, entertainment and gaming? Two Orion wearers able to play a multi-player game, have their own AR versions of Messenger (or WhatsApp) and be able to jump right into an AR ready document editor or an email app, could just be the ticket. We’ll just have to see how the hardware, and subsequently the software experiences, evolve. We’ve surely not heard the last of Meta’s Orion pursuits. Meta iPhone-esque moment? Time will testify.
Meta AI’s Imagine feature is set to find space in your Instagram and Facebook feeds, Stories and profile. That could be problematic.
There was more at Meta Connect.
- The Ray-Ban Meta smart glasses can now remember things for you. To-do lists, shopping lists and so on. In an year when AI wearables have largely flopped (looking at you too, Humane AI Pin) for more reasons that just the pricing, Meta’s efforts with the smart glasses have proved to be a success. Now, Meta AI too will be more responsive when called upon via the sunglasses. There are also plans for real-time video processing to better understand the world around you (for now, the situational awareness is restricted to photos), but that’s not expected immediately.
- You may have opinions about AI and its usefulness, but chances are, you’ll want to give Meta AI’s voice conversation feature a try. It’s coming soon as an update within WhatsApp, Instagram, Facebook and Messenger. The AI voices will include celebrity voices including Awkwafina, Dame Judi Dench, John Cena, Keegan Michael Key and Kristen Bell (this being a surprising deal to crack, considering her opposition over the summer to Meta using her data for AI training).
- Now this may be problematic. Meta AI’s Imagine feature is set to find space in your Instagram and Facebook feeds, Stories and profile. “You can now imagine yourself as a superhero or anything else,” is Meta’s pitch, adding that “Meta AI can also suggest captions for your Stories on Facebook and Instagram.” Make of it what you will.
- “And we’re testing new Meta AI-generated content in your Facebook and Instagram feeds, so you may see images from Meta AI created just for you (based on your interests or current trends). You can tap a suggested prompt to take that content in a new direction or swipe to Imagine new content in real time.” I’ll just leave this here, for you to make your opinion on.
Can Meta build an AR platform, the kind that doesn’t exist even though Snap and Apple have been trying with AR hardware and the developer ecosystem?
More serious matters for a moment. Just a couple of months after Meta released the Llama 3 AI model, there’s an update. Llama 3.2 becomes the first open-source model capable of processing both images and text. Remember I’d spoken earlier about a potential AR platform? Llama 3.2 could be useful for developers for augmented reality content, analysing photos and videos as well as real-time context of the world via video. Interestingly enough, there will be a smaller version of this model, optimised for mobile devices.
MARKET (REALITIES)
You’d have noticed I’ve been closely tracking the AI PC / Copilot+ PC developments since things materialised on that front over the summer of 2024. Now that Intel’s gotten into the ring in the past few weeks with their first true AI PC machines finally going on sale thanks to the likes of Asus (I’ll be reviewing that in detail) and HP to follow, we now have the troika of choice—Qualcomm, AMD and now Intel. This provided a timely backdrop for my conversation with Asus India’s Arnold Su.
A few interesting observations emerged, with Su detailing something I’d suspected for a while now—AI PCs aren’t exactly new, because even the previous generation chips had some level of processing capabilities for these models. Just that the new definition that everyone has agreed to means 45 trillion operations per second (or TOPS) as the baseline for a neural processing unit (NPU) to be called an AI PC.
My profession puts me in the rather unique position of receiving feedback from friends, family and readers about their experiences with service centres. As they are, in our country. Su detailed to HT something that Asus is working on, the sort of thing that we haven’t seen in such vivid detail in India’s PC ecosystem. That is, leveraging retail points to become a link in the service chain. Useful, because it gives users much more convenient access, and saves companies the dual investment of retail and service in same or adjacent geographical locations. As a Asus laptop user, I’ll be able to drop it off at a nearby store, and the store will be responsible to take it to a service center for repairs and collect it back once that’s completed. The user again has to go to that store for collecting their laptop.
In theory, it sounds very neat. However, on-ground implementation will be far less simple. Have you ever tried to reach out to your nearby Croma or Vijay Sales or Reliance Digital outlets for any level of service (pre-delivery, post-delivery or subsequent repair or service)? It is akin to seeing a brick wall, and wilfully deciding to introduce your forehead to it. Within this paradigm of limited responsibility that often defines retail mindsets in our country, Asus has their task cut out.