In Full...
If you’re familiar with the Apple App world you may sometimes have heard someone refer to an app having been “Sherlocked”. This basically means that Apple has decided to add that App’s core feature to their main operating system, be it iOS, MacOS or iPadOS. It does not usually end well for the original app.
Sherlocked?
Way back, in the mists of time (1998), Mac OS came with an app called Sherlock. It was originally an extension to Finder that allowed users to search not just by file name, but also within the contents of files.
Sherlock did more than that though. It also allowed users to search on the Internet, without using a browser. You could tap in, say, “Sherlock Holmes” and select multiple search engines — AltaVista, Excite and Lycos of course — and it would pull up the results. This was quite a cool thing to be able to do.
Sherlock became very popular and the next version of the software let users use the same system-based, universal interface to search the web via different channels — Entertainment, Pictures, Shopping, News — for whatever the user wanted. An entrepreneurial company then built another piece of software: Watson. Watson was similar to Sherlock, but it allowed you to search on specific sites; ebay, Amazon and more. Not only that, but it also allowed you to buy things on these sites—it was sort of the first universal e-commerce app. This was a very cool thing to be able to do!
Apple, keen to improve its own tool, then added to Sherlock the ability to… search on specific sites; ebay, Amazon and more. Not only that, but it also allowed you to buy things on these sites. Ah.
Many websites rushed to develop their own channels or extensions for Sherlock so that they could appear within the app. Sherlock was a success and Apple was very happy.
The makers of Watson though, were not happy. Apple, at a time when they still responded to these things, argued that the features were an inevitable evolution of Sherlock. Watson thus became the first app to be “Sherlocked”.
Well, that’s a relatively neat origin story. The pattern remains the same to this day. Every year, Apple improves their own software by adding features that until that point third-party apps have solved, and websites like TechCrunch get a good annual article to write.
Sherlock the app though, eventually disappeared, tumbling to its death in 2007.
There’s An App For That
At around the same time that Sherlock disappeared, the tech industry and the businesses which came to dominate the following decade, were starting to understand the importance of analytics data and the role that their sites’ designs played in both user retention and engagement. As their strategies improved, many sites closed down avenues through which users could engage via third parties like Sherlock. It became increasingly important to “own” the relationship with the consumer.
The arrival of Apps in 2008 saw companies realise that these offered an even more direct means of owning that relationship. We saw websites push users towards installing apps on their phones, partially so that there was an ever-present reminder of what to tap on to find what you’re looking for but also to gain access to those all-important notifications. Third party tools like Sherlock were no more, but now we lived in the future, where we could wake up in the middle of the night to our phones buzzing with an important offer on socks.
We also saw the emergence of the “big players”.
Sites like Amazon had realised that the real play wasn’t just in being a direct seller, it was in being the aggregator of all sellers. For Amazon, this, in combination with Bezos’ maniacal focus on customer experience, led to the Amazon we know today: a massive and comprehensive product inventory, free next-day shipping and one-click checkout. The Amazon experience is so well executed that half of all users don’t start searches for products on Google, they go directly to Amazon.
And this has been the status quo for many years now. We have a handful of incredibly well executed aggregators that own the various different sectors. Aggregators of this sort, as you may be familiar, dominate the internet. All of the biggest players in every space are in some form or another, a type of aggregator — Amazon for retail, booking.com for holidays, Netflix for entertainment, uber for taxis — these businesses don’t necessarily provide the end product from their own supply, but they do provide access to the service in the most convenient way.
The key for aggregators is their direct relationship with the user, which Ben Thompson describes as “the linchpin on which everything else rests”. As long as the aggregators can maintain this direct relationship and ensure that their service provides the most convenient way for users to access the services, they will maintain their dominance.
AI Personal Assistants
In January, I wrote about AI based Personal Assistants, sparked by emerging products from companies like Humane and Rabbit. Those hardware devices, as I’d suspected, were DOA, with MKBHD describing the Humane Pin as “The Worst Product I’ve Ever Reviewed”. But does this mean the idea of an AI personal assistant is dead? Not quite.
The reason that humane and rabbit stood out was because they were a new class of hardware; if they’d just been apps then they’d have received relatively little airtime. And it turned out that rabbit was just an app, in a fancy new housing. The problem however with being “just an app” goes beyond marketability.
The challenge with creating an effective AI personal assistant is access. It needs to interact with everything on your device. This level of integration is only truly possible for device manufacturers like Apple, Google, and Samsung. Apple, in particular, seems to be embracing this opportunity with Apple Intelligence and Siri.
While headline AI features baked into the OS—improved contextual search, writing tools, smart replies, summarized notifications—are impressive, Siri’s real power as an AI PA lies in its ability to take actions within other apps, potentially without direct user interaction.
At first glance, this seems like a win for everyone. Imagine Siri popping up with a timely notification to book an Uber when it recognizes you have a dinner reservation in 20 minutes. It’s like ultra-targeted advertising for free. But there’s a catch: Uber (and companies like it) risk losing their direct relationship with users.
The Aggregator’s Dilemma
This puts these businesses in a tricky spot. Users adapt quickly to new paradigms. If it becomes normal to rely on Siri (or its Android equivalent) as the first port of call for actions like booking a car or a table, apps will need to provide these actions to Siri or watch as competitors do so. However, this loss of direct interaction means services become more interchangeable, with the customer relationship shifting from the app to Siri. This is the risk that I wrote about in AI as a Personal Aggregator.
The big aggregators stand to have the linchpin Thompson talks about snapped – if the first place that a user goes to find what they need is Siri, then they lose one of the structural pillars on which their market dominance rests.
This shift towards AI-driven personal assistants as primary interfaces carries profound implications for the tech ecosystem. For users, it promises unprecedented convenience—imagine a world where your device anticipates your needs, seamlessly integrating various services without you having to juggle multiple apps. Need a ride? Siri books it. Forgot to order groceries? Your AI assistant has already compiled a list based on your usual purchases and dietary preferences.
However, for app developers and service providers, this new paradigm presents both opportunities and challenges. On one hand, integration with AI assistants could provide access to a broader user base and more contextual usage. On the other, it risks commoditising their services and weakening brand loyalty.
For the tech giants developing these AI assistants, it’s a potential goldmine. By positioning themselves as the gatekeepers of user interaction, they could gain unprecedented insight into user behavior and preferences. This data could be leveraged to further improve their services, creating a virtuous cycle of increasing utility but also dependence.
The Final Chapter
Of course, the aggregators won’t go down without a fight. They’ll integrate, they’ll innovate, they’ll probably coin some buzzwords about “AI-enhanced user experiences.” But at the end of the day, they might just be delaying the inevitable. After all, why juggle a dozen apps when one AI assistant can do it all for you?
In a twist of digital déjà vu, we seem to be circling back to a universal interface for accessing third-party services on our devices. Perhaps Sherlock wasn’t dead at all, maybe it just has a new name: Siri.>