Excuse my crappy painting skills. Appears like I didn’t become a painter for a reason…

The Last Touchpoint: The Future Of The Web As Imagined By Google (According To I/O)

The internet as we know it today is changing. Google’s new emphasize on intelligent, conversational interfaces, as embodied by Google Assistant, may prove to be a catalyst of those developments. The most significant one is the modular web, a radically different way to experience the web —some might wish: Via a single touchpoint.

I haven’t been as techcited as I am since the announcements made at Google I/O. For those of you who care, you can watch the full keynote here:

So, what’s the big deal?

Well, while only time will tell whether or not Google will succeed with its newest endeavors, what they presented this week at least has the potential to be regarded as a landmark in the internet’s development (though the term ‘internet’ might actually be misleading when trying to understand what’s happening. That’s because we associate it with the known architecture that consists of web browsers, pages and so on which is exactly what we are moving away from).

In this piece I am going to first summarize which technology and products Google presented at its annual developer conference. Then, I will put it in the bigger context of where the web is moving. And finally, I’ll talk about the likelihood of Google succeeding.

Google’s Focus: Artificial Intelligence, Machine Learning and Assistive Technology

Nowadays, everybody is talking about artificial intelligence, machine learning, neural networks, natural language processing, bots and so forth. So it’s no surprise that Google is working on and doing research in all these fields (or quite the opposite, as Google has long been known as a driving force in those fields, see the AlphaGo story). They even incrementally introduced certain features into their products — for instance smart replies in Gmail, voice search etc.

However, Google has now apparently bundled his diverse — and in many areas leading — technology into a product (or service. Or tool. Or brand): Google Assistant. In their keynote, they showcased what it can already do today. And it was rather impressive (though all we know about the technology is from a staged, prepared product demonstration). It was the kind of moment when a future, that you knew was coming, actually manifests itself and surprises you regardless of you expecting it.

As I assume that you have by now heard about Google Assistant’s presented functionality I won’t go into a lot of detail here (I’ll provide some links to good recaps at the end in case you want more details. Or simply watch the video).

Google Assistant is basically a bunch of the aforementioned technology — AI, natural language processing & machine learning — bundled into a tool that lives in the cloud. Google CEO Sundar Pichai describes the underlying principles as:

  • Conversational
  • Get things Done
  • In your World

That is, you can have actual, context-sensitive conversations with Google Assistant. Those conversations aim at solving your problems (and: meeting your demands) in the real world. You ask the assistant what movies are on tonight and receive an answer in the form of a list of movies that play in your town. You go ahead and ask if one of the suggested movies is ‘any good’ and it provides you some reviews. Then you tell the assistant to order tickets and it does so for you. You get the idea. Of course, the system is self-learning, so it’s going to become more useful over time as it learns to understand your preferences. Or as Pichai put it:

“We think of it as building each user their own, individual Google”

As it appears right now, there are going to be three ways how you will be able to interact with Google Assistant:

  • The Google app
  • Google Home, itshome device, basically the equivalent to Amazon’s Echo
  • Allo, Google’s soon-to-be-launched messenger app

The common denominator: All of them are voice-controlled interfaces that allow you to direct your questions at Google Assistant (in Allo you can obviously also correspond in writing).

What’s noteworthy about Allo is that the assistant will also integrate into your conversations with others. That is, it is not limited to 1-on-1 interactions.

So much for what Google presented.

Where We Are Heading: The Modular Web, Assistive Search, Conversational Commerce

Taking a look at the overall picture that presents itself after Google I/O I can’t but say: It’s quite a biggie.

For one, it’s a culmination of many trends we have witnessed in the web’s and — more broadly — tech’s development. We have been talking about and witnessing several companies leap into artificial intelligence, conversational assistants (aka bots), IoT/smart home/connected devices and machine learning for quite some time now. It’s not news, that the web is moving away from a page-based approach to a more free-flowing, user-behavior-centric distribution of/access to information (Facebook Instant Articles, Google Cards, Medium).

This part of the story is universal and not specific to Google. However, Google is a player that might be capable of driving and catalyzing these developments at scale. And since it appears that the search company (probably no longer the right term) is putting a lot of emphasize (translates to: resources) on doing so, it’s time to take a look from a broad perspective.

While all the information I have on Google’s new products right now is from hearsay and not actual use, this only has potential implications on the short-term success of Google’s new products. It doesn’t really affect the broader points I’m going to make. So without further ado, the three trends that emerge from Google’s announcements are:

  • The modular web
  • Assistive Search
  • Conversational Commerce

Let me depict what I mean with each of them and what the respective implications are.

The Modular Web

The web is becoming exceedingly modular. For the longest stretch of its existence, we were used to the internet as something that consists of many singular websites that are interconnected mostly via hyperlinks. In order for us to access a certain information, we had to visit a website which provided it. In contrast, in the modular web every piece of information is a data point that can be accessed from different frontends or user interfaces.

As I stated above, this modularization can already be witnessed. If you type a term like “Weather Munich” into Google, you get the information delivered straight in your search results. The underlying data is provided by weather.com but you don’t have to visit their site. Facebook Instant Articles does the same with news & reporting. Many people think that bot technology is such a big deal for the same reason: Instead of interacting with all those different entities across the web, the idea is that you use the messenger as a single interface via which you interact with an AI that does the hard part for you.

Developing this one, primary interface is seen somewhat like the (next) holy grail of internet technology. It gives you a HUGE amount of power and control over users, markets and even information in and of itself (that’s why regulation will almost certainly play a role at some point in time). Thus, many companies are currently working on solving this equation. Therefore, regardless of who is going to come up with a widely adopted solution, it’s worthwhile to look at the implications. They are everywhere. I’ll give you random list of points that come to mind — without raising any claim to completeness — and elaborate on some:

  • “Web design” (don’t think of Photoshop but system and service design)
  • The internet’s architecture
  • The economy
  • Business models
  • Media/journalism
  • Finance and transactions
  • (User) behavior
  • Opinion building processes in society (think of the facebook news scandal — times x)
  • to be continued

Let me dive deeper into a few of them.

“Web Design” / The Internet’s Architecture

I described the web as it (mostly) is today above. In the modular web, however, we don’t need to optimize our “design” for visitors of our page but we have to optimize our information/data for compatibility with different user interfaces. This sounds simple but actually alters completely how we think about “web design”. It’s so radically different that I use quotation marks. All of a sudden we don’t “operate a website” anymore but broker information and data — processed to different degrees.

To seize the cinema ticket example from above: Today, if you are in the business of selling cinema tickets, you probably aim to offer relevant data along the customer journey on your site. You offerinformation about which movies are playing, you might offer user and/or professional reviews (more likely when you are an aggregator than when you are the cinema as the former doesn’t have to care about filling every hall), you have information on the availability of tickets as well as their price and you provide an interface for transactions so the user can eventually buy the ticket.

It’s actually a good example because even in today’s web we find different degrees of modularity. A cinema might create and offer most information itself, except maybe reviews which it aggregates from another site. A site like Fandango is already in the business of modularity as it aggregates data from different sources/entities. The movie synopsis come from Rovi, the price data from the different cinemas (or, I suspect, rather from another company that collects it) and so forth. So, in my above terminology, Fandango is a user interface for buying tickets.

Now, imagine one interface that provides this kind of functionality — only that it does it for everything (okay, likely an exaggeration for at least the near future, as I don’t see, for instance, building enterprises buying excavators via instant messengers for a while. But you get the picture).

Not only does it potentially marginalize the value of having a nicely designed (current meaning) website — to the contrary it is a natural driver of increased specialization. In the website world, it makes sense to offer a lot of information relevant to your user at a central place (including aggregated content) as it creates a superior user experience. However, this no longer applies in the world of a (few) central, intelligent user interfaces aka assistants— except when you are the assistant (I’ll explain later why there’s only room for very few of them. In brief: network effects and user experience).

Instead, the assistant — and therefore the user — is indifferent to where a piece of information comes from. The assistant aggregates, processes and presents it anyway. It is, in an ideal state (as envisioned by Google & Co., not by me), the single touchpoint with the user. Everybody else is only a data provider. At this point my initial reflex would be to say ‘is reduced to’ but that is not necessarily the case. It’ll take some time to wrap my head around this but I believe there are ways to create a lot of value by providing the right kind of highly valuable data. Plus, until then the short-term complexity is going to be so high that it’s going to take at least a couple more years before the impact will be felt in most markets.

The Economy

I will keep this brief (for now) because some time will have to pass before we see any large scale impact anyway and many variables are unknown. Or, to put it bluntly: It’s just too much guesswork involved right now. So I will simply share the broad implications and coherences I can envision while highlighting that I’m not necessarily confident in them — in sports you would call them my hot takes (I share them anyway to stimulate a discussion and gain feedback/perspective).

1: Today we analyse the tendency to create monopolies inherent to platform business models. The assistant that will eventually create wide-spread adoption is going to be THE platform. You could also think of it as a meta-platform of sorts. A platform of platforms. Deeply embeded in everybody’s life, it has the potential to be the most-used touchpoint with every business and service. This of course comes close to market domination — in (almost) every market. It is therefore super likely that regulation will kick in. Also, its possible that ownership models of platforms in general and of such a platform in particular will change, probably rendering them as public goods at some point in time. (That’s a point Ben Thompson & James Allworth discussed recently in an exponent.fm podcast — I’ll provide a link to the episode later as soon as I find it).

Again: That’s a lot of guesswork. Everything might develop differently as the future is path dependent and things might change in the meantime.

2: Less big, more small. That’s a direct consequence of the specialization aspect I mentioned above. Combine it with the general development in automation and decentralized system design and we might be heading into a future where big — as in controlling the complete pipeline — is simply not the most efficient state in many industries. Of course, this might be highly industry-specific. Services and businesses highly driven by information/data are likely to go in such a direction first. But as we get increasingly better in organizing decentralized systems, I can envision futures where even complex manufacturing work is modularized. Whereas the former is likely to be driven by a general life assistant like Google’s, the latter will need more specific AIs.

Business Models

The impact on business models is going to be felt quicker in some areas than in others. As stated above: The more your business is built around information and services, the quicker assistants will impact you. Also, as the currently introduced assistants target people in their private lives, b2c businesses are generally more likely to be impacted at first.

I hinted at many aspects of business model change in my description of the modular web, so I’ll just briefly recap those here:

  • Increased specialization as described above directly influences what kind of business models & company structures are most efficient. Often, those are likely smaller and more narrow in scope than before.
  • Whatever your company actually does, in order to compete in the modular web you have to think about your business in terms of information and data. What different types of relevant data do you actually provide and how can you modularize them so they can interact nicely with assistants? This is a tough challenge as it is a totally different way to think about a business compared to what’s commonplace today.
  • If nobody visits your website anymore, you can’t monetize page visitors — be it via advertising, direct sales or subscriptions — you need to create revenue streams that work in the realm of assistant-based conversational commerce. This is likely going to be a collaborative effort between the owner of the assistant and the actual producers/providers as they are co-dependent.

And finally: Everything as a service. I borrow this phrase straight from Ben Thompson who wrote a fine piece about the platform economy’s tendency to turn business models that formerly sold products into services. We see it in software and we see it with the sharing economy (a euphemism for turning under-utilized but usually owned products into pay-per-use services). This development will likely be accelerated when there is a meta-platform at scale.

Media & Journalism

I won’t say a lot on this topic as Jeff Jarvis wrote a nice piece on it. What he calls ‘chunky content’ is basically his take on the modular web from a media standpoint. His analysis is spot-on. He (and Carrie Brown, who he quotes) basically say about Journalism what I said more broadly for businesses above: It needs to think about it as a service, not a product.

Combine this with the other points I made above and you end up with a media world that revolves around way smaller entities than today’s media corporations — probably as small as singular voices from well-established personal brands — who’s content is organized and delivered by intelligent assistants.

That’s also not a new development by the way, as this is what we see today with sites like Medium or how being Bill Simmons is way closer to being ESPN from a brand value perspective than maybe even only 10 years ago. The only difference is in scale and speed — and as I said: Any industry that is built mostly around information is likely to be affected first by intelligent assistants. As that’s the definition of the media business model (at least as understood today in the internet world; there might or might not be some news publishers who believe otherwise but they will be extinct soon [I usually avoid generalizations but in this case I am happily generalizing]), we will continue to see a lot of change there.

Opinion Building Processes In Society

We just had the Facebook News scandal. We have been talking about filter bubbles for a while now. The dangers and implications have been tackled by many people. Why I mention it, again, is scale. In a world where we have an intelligent assistant as kind of our life’s operating system that is largely responsible for us accessing information, this operating system has a HUGE power and responsibility.

Yes, it’s nice that Google’s code of conduct contains the well-known “don’t be evil” slogan but that’s hardly an insurance. At least their track record in providing unbiased information is better than Facebook’s. But there is not only intentional, man-made bias but also tech-made bias, as showcased in — surprise! — the basic logic behind the google algorithm: The most popular (as, simplified, measured in backlinks) information is regarded as most relevant, therefore placed at the top of the results, therefore becoming more popular and so on.

Combine this with our psychological tendency to prefer content that reinforces our believes and this becomes a very interesting puzzle to solve. At least if you believe there is value in things like a broad discourse among society, a balanced perspective, minority opinions and so forth. I’m not saying this can’t be solved but it’s about time to address this issue emphatically.

(I’ll keep all the other points in my list for another time and/or for people more knowledgeable on the topics).

Assistive Search & Conversational Commerce

If you made it until here, I’m glad to inform you that the latter two points and their respective implications are mostly already addressed.

So I’ll keep it brief.

Assistive Search

Search as we know it will be gone, replaced by a superior, AI-based user experience. This won’t happen tomorrow but it’s clearly the direction Google is going and it makes sense for them to do so. Chris Dixon wrote a brilliant piece on the Internet Economic Loop. He writes:

“ Think of the internet economic loop as a model train track. Positions in front of you can redirect traffic around you. Positions after you can build new tracks that bypass you. New technologies come along (which often look toy-like and unthreatening at first) that create entirely new tracks that render the previous tracks obsolete.”

Search in its current form is increasingly marginalized because users ‘find’ things on the web differently — that is, via social (so actually a lot of the time the things find them). Search works well when you have a specific need. But overall, users spent way more time in the internet’s social environments.

In this regard, Google Assistant and the products around it are a potential game changer — provided Google is successful which remains to be seen (more on this later). They aim to be so deeply embeded in our everyday lives that I just called them ‘the operating system of our lives’ a few sentences before.

As Google puts it, they focus on getting things done in our real lives. If they can deliver on this promise, conventional search will hardly matter. This is a big deal for everybody who is in the business of reaching his users/customers/audience via the web. Today, that’s basically every business (to varying degrees, obv). It’s for this reason that it might become highly relevant to think about the aspects I wrote about in the section on The Modular Web.

Conversational Commerce

I touched on this in the business model part above. Basically, when users start buying stuff in a conversational manner using the interfaces to intelligent assistants at scale, we must design our business models accordingly. I call that conversational commerce. And it has some very real potential to provide a much better user experience than established ways of shopping, particularly in categories that aren’t optimized by very effective platforms yet.

Google launching Allo highlights that they think about it the same way. It’s not because they want to provide us with an excellent user experience in messaging — the potential gain would be way too small compared to the necessary investment it will take to successfully enter this already well-catered market — but because they think it might turn into a significant revenue stream (I have no inside information to validate these claims; these are plain assumptions based on common sense and a certain trust in Google’s smarts — and ability to learn from past mistakes, say Google Plus or Wave).

As outlined earlier, this won’t affect every industry & business at the same time, so different companies are on different time frames. But history shows us: It’s never too early to think about changes appearing at the horizon.

Can Google Actually Pull It Off?

So far, I have been talking in more general terms about the current developments Google wants to accelerate and play a significant role in. I think it’s relatively safe to assume that this is the direction we are going in. Now, the interesting question remains: What will be the time frame?

Alas, I won’t give you an answer as I don’t have one. But what I can do is take a look at the likelihood of Google successfully establishing its new products built around Google Assistant in the market place. This is going to be a decisive factor when it comes to timing.

Let me frame the issue first. To me, success will highly depend on a few variables:

  • Technology in AI, machine learning, natural language processing
  • Capabilities in accessing, organizing, interpreting and delivering information/data
  • Product-market-fit of the assistant’s interfaces / initial use cases
  • Ability to achieve scale, quickly
  • Available resources to drive further development

Based on those assumption, lets drill down deeper.

Technology in AI, machine learning, natural language processing

According to Pichai, Google owns the world’s best natural language understanding (he makes a point of stating that this is especially true when it comes to follow up querries). If we are to take his word at face value, they are “an order of magnitude better than competition”.

As I am no expert in this industry, this might or might not be true. There might be a small company, maybe a university spin-off, that is even better (mind you that “competition” might only refer to big players in tech). I doubt it, but even if there were: It’s basically beside the point.

Because even if it were, Google could simply buy it. Technology is important, but its not going to be the decisive factor. A well-known software industry principle applies: The technology must be good enough. And when the first available Google Assistant products work only nearly as good as the demos at I/O make me believe, that certainly is the case. I at least haven’t seen anything better in a consumer product. Plus, the fact that in Ray Kurzweil an industry legend is overseeing the natural language processing team at Google gives me confidence in their capabilities.

Another important factor since we are talking about a global market place: Languages. I didn’t find any reliable numbers on how many languages Google’s speech recognition works well enough in to provide useful, voice based assistants but Google is working on the issue for a few years now within its Android ecosystem. Thus. I’m confident that they are among the leaders in this category as well — a major asset for scaling it globally.

When it comes to AI and machine learning, the same logic of good enough applies. I don’t know if IBM’s Watson or Google’s artificial intelligence(s) are “better” — it might likely be comparing apples to oranges anyway— the fact is: Google is pretty good in this field as well. AlphaGo, Googles Go-playing AI developed by DeepMind, beating one of the world’s best human Go player in Lee Seedol 4:1 — about 10 years before most experts expected this to happen — is a good indicator.

In sum, I think Google is well positioned technology-wise to be a major player in the intelligent assistant market.

Capabilities in accessing, organizing, interpreting and delivering information/data

In my opinion, this category likely makes the strongest case in Google’s favor. Remember what I described earlier as The Modular Web. In order for those intelligent assistance to be truly useful to people, they need to be able to draw all kinds of information from around the web, process and make use of them, deliver them to the user in a useful format and allow for further interaction with it. And all this, ideally, within a fraction of a second.

Guess who is doing this right now at an immense scale, millions of times per day? Right, Google. Of course, delivering a truly conversational experience based on voice is another way to apply its capabilities and comes with its own challenges but at its core there is nobody out there who is even remotely close to Google when it comes to knowing, analyzing, modularizing and delivering information, particularly at that scale. To me, this is why I suspect that Google Home is going to vastly outperform Amazon Echo, which is the closest, most comparable AI interface product out there.

Google owns this category. A category that is key to building a truly useful product and great user experience. This is not to say there won’t be room for other solutions. Particularly in the early phases of market maturation we are likely going to see several players that focus on different initial use cases. So there might be ways to outperform Google’s data handling capabilities in some niches. Particularly when it comes to special applications and assistants specific to more complicated cases. However, there is a large advantage in the consumer market to be the one-size-fits-all solution, simply because of convenience. And Google is the player best equipped for this future because of what it already does today.

Product-market-fit of the assistant’s interfaces / initial use cases

This is going to be an interesting one. Ben Thompson wrote a great piece on Stratechery, highlighting why he is skeptical when it comes to this. And indeed, when looking at Google’s past we find different examples of failed products and untimely entries into new markets.

There is the case of Google Wave, a product that was launched with much fanfare but failed to meet any actual demand (it is now called Apache Wave and lives at the edges of the internet). And there is, most famously, Google Plus which was Google’s attempt to break into the social networking market and compete with Facebook (and probably Twitter at the time). Even though it is still being continued (in a stripped down version) it is nowhere close to relevant, at least in comparison to Facebook and even Twitter. Last year, some former Google employees talked about their perceptions of why it wasn’t successful.

To me, the most relevant mistakes (both are also mentioned in the article) were bad timing and a lack of product differentiation — basically Google didn’t understand the market and its dynamics. When it launched in 2011, Facebook had already achieved a dominant position. Network effects where in full effect. By then, the switching costs were pretty high. Thus, in order to compete you would either have to offer a far superior user experience OR a truly differentiated product (that’s why Snapchat or Instagram which came later managed to grow next to Facebook). Google Plus did neither, even though it provided some interesting new features.

So, yes, there is some history we can point to where Google did get it wrong. But I’d like to argue that the intelligent assistant and voice-controlled interfaces market is something else entirely and Google is in a way better position than it was in those cases.

An Emerging Market

Wave didn’t cater to any particular market — it was a fancy idea and cool technology but not driven by any demand. Social Networking was already maturing when Google entered it. Neither is the case with voice-controlled, intelligent assistants (I’m framing the market in this particular way on purpose, even though consumers might not).

Coming back to my scribble. At the surface, we have to talk about Google Home and Allo as two separate products. One is a voice-based home controller (aka Google’s Echo) and the other a messaging app. But from Google’s perspective they should merely be two different interfaces or user experiences that create lock-in to the Google Assistant ecosystem. Thus, at least from a broader perspective, the individual products’ success should be less relevant to them than the overall goal of getting people to fall in love with interacting with Google Assistant. What will finally be the preferred method to do so might not even be developed yet.

Of course, the picture looks different in the short-term. Right now, it’s all about creating said interactions and getting people used to them while continuously increasing the amount of (perceived) value they create. The best and easiest way to achieve this is by driving adoption of the currently available interfaces. So let’s take a look at Home and Allo respectively.

Google Home or The Fight For Dominance In The Next Operating System Generation

Depending on your perspective, Amazon’s Echo and Home might only be somewhat useful — basically toys for grown-ups. Since I haven’t used any of them myself, I can’t really tell. You can find some videos on the web made by actual Echo owners, like this one:

Judging from this I probably wouldn’t buy one though it is actually somewhat useful. When it comes to Google Home, we have a by all means impressive but of course biased product demo, so we don’t really know.

They both are certainly neat devices. But they are also way more. They are the first in (as I suspect) a row of devices that carry what is going to be the next generation of operating systems into our living rooms. The voice-based interfaces that connect us with intelligent assistants who live in the cloud and draw from the entirety of the internet are the next playing field for tech companies to build a platform that drives our interactions with technology.

We moved from desktop operating systems to mobile operating systems. Also, we have something like operating systems built on operating systems — the browser is a good example. You also might consider Facebook as the social operating system. A point both, Dixon and Thompson stress quite rightly: You always want to be the one closest, most relevant to the user in those multi-layered structures (maybe that’s why Google sold more Chromebooks than Apple sold Macs in Q1 2016: The browser is more important to users than the OS).

And what would be more relevant and close to the user than his intelligent assistant who aggregates and organizes all information for him? Exactly, maybe only internet implants.

Even though we are still far from the more drastic scenarios I described above, I think this framing is how Amazon, Google and likely others look at the market that Echo and Home just opened.

Let’s get more specific to get a better grasp of the market and Google’s situation within it. As of early April 2016, Amazon has sold over 3 million Echo devices in the US. It became widely available at the end of June 2015 so we are talking roughly three quarters. For comparison: According to Statista data, the iPhone sold approx. 3.7 million units in its first three quarters of existence. So both are in comparable figures when it comes to units, though the iPhone was obviously more expensive. Still: This definitely constitutes an existing market with real demand an lots of potential.

Structurally, this is much more like the smartphone market back in 2008, when Google entered it with the first Android phone, roughly one year after Apple launched the iPhone. So there is some experience with similar situations in the company (even though this time around Google opted to built his own hardware, an interesting choice in its own regard).

Remains the question whether Google will be able to offer a better or at least differentiated product. We won’t be able to conclusively answer it before we actually get our hands on it. However, there is reason to be optimistic because of Google’s aforementioned capability to handle information at scale. Outside of the commercial use cases — which are important to the business side of it but not the most relevant selling proposition for the users — they are way better positioned to deliver a great experience on this end than Amazon is.

To summarize: If Google doesn’t actually mess up the product side, we are in for an interesting clash of two giants in today’s tech industry.

Allo and The Messaging Market

The messaging market is a different animal. It resembles the Google Plus situation more closely. It’s a way better served market and one that is a) diversified internationally and b) shows enormous network effects (the messenger is only as useful as the number of your contacts using it). Switching costs are much higher.

So for Allo to be a success, differentiation is key. What I’ve seen from Google I/O I’m rather indecisive. Of course, the integration of AI/Assistant into the messenger is very interesting from a technology perspective and it might create a better user experience. The problem, however, is that users likely only learn that when they use it. It remains to be seen if this feature alone can drive adoption outside the hardcore tech community. There are reasons to be skeptical. For now, without ever having touched it, I’ll take the easy way out and say: We shall see.

The Core Products: Search and Mobile

Let’s not forget the integration of Google Assistant with the core products, search and mobile. As Pichai noted, one in five search queries are via voice as of now. This number surprised me — a typical case of selective perception because I hardly ever use it. Plus, 50% of queries happen mobile.

So how and how well the progressive transition from search as we know it to assistive search will happen is likely THE decisive factor for the success of Google’s endeavor. They have many assets in their favor: They are the number one service globally that people use when they are looking for information. Additionally, Google owns the most popular mobile OS. Again: Globally. Search and Android are closely integrated and Google Assistant will likely become an additional layer on the phones.

I stated above that the main interface via which people interact with their intelligent assistant might not have been developed yet. The opposite might be true as well, at least in the near future: It might simply be their phones.

So, to sum it all up: Market-fit is a tricky one as we first have to understand the market we are operating in. Then, we have to look at it differently for the short- and mid-term. While we don’t have enough information to talk about the latter, the former is a mixed bag. Nevertheless, I’m rather optimistic overall that Google will become a relevant player in the intelligent assistant and voice-controlled interfaces market because at least some of their interfaces will work out short-term. Also, timing appears to be about right as the early success of Echo & the adoption of voice-based interactions with Google demonstrate.

Ability to achieve scale, quickly

This one is rather self-explanatory. Once you created product-market-fit you must be able to scale it. Google is extremely well positioned to do so for all the reasons stated above. If they can create valuable core interactions with their AI interfaces, they are well positioned to create the necessary ecosystem around it.

One core component of this has already been initiated: Creating an API that is open to Google’s (well-established) developer base. In doing so, Google won’t be the only one having to come up with use cases and more valuable interactions. This is an important asset. Also, they have the technical infrastructure — probably only matched by Amazon — to scale it quickly and without having to expect major outages even if it would suddenly start growing exponentially.

Summary: Check.

Available resources to drive further development

Again, this is an easy one. Money, people, smarts: Google has it all. And judging by what I saw from I/O, I assume they are more then willing to put a lot of it into this. I will spare you the details, as I figure that anybody who read that far is capable of googling it her/himself.

Summary: Check.

Google is in pretty good shape

Going into writing this, I was more ambiguous about my assessment of Google’s potential to drive home some short-term successes based on what they revealed at I/O. While a lot can happen underway, right now I’m rather optimistic that we will see increased use of Google Assistant technology in the not-so-distant future.

Google is well positioned to enter the market and timing appears about right. Plus, they have a lot of incentive to direct a fair amount of resources towards being successful as Assistant is potentially capable of solving the problem of being increasingly marginalized in the internet economic loop (to borrow Dixon’s term again).

These are interesting times!

Wrapping It All Up

If you made it until here, first off let me thank you for your attention. You can clap yourself on the back as you are a living rebuttal to the claim that content on the web needs to be short (say, 600 words or so) or nobody will have the attention span to digest it. I never believed that and now I know that I’m not alone.

That being said, let me recap aka give you the


  • Google’s 2016 I/O might be regarded as a landmark in the web’s history when looking back at it from the future.
  • The search giant announced his plans to tie together its capabilities in machine learning, natural language processing and artificial intelligence as well as in information handling (its core business) in order to enter the market of intelligent, voice-controlled assistants.
  • On a general level, this is only the latest (and to this day biggest) project in a bigger context of developments around tech and the web. Regardless of who will eventually succeed in the evolving market, it will go along with some big implications:
  • The modular web is the most impactful one. The term describes the future of a web that’s very different from the one we have today. It’s not based around websites but pieces of data that are being accessed via interfaces humans use to communicate with intelligent assistants. Those, in turn, aggregate and process the data from different sources and in doing so aim to create the last and only touchpoint we use. This in turn will influence many domains from web design to the economy, business models, journalism and the society more broadly.
  • The other two implications are assistive search — a very different way of accessing information — and conversational commerce. Both are closely related to the modular web.
  • After looking at those general developments, I analyzed Google’s potential to be successful in this endeavor. While the devil is in the detail, the overall impression is that Google is positioned pretty well to make it all happen.

Some remarks I didn’t want to put into the main text

The end of the free web?

It’s an issue that is currently being discussed and really concerns me. The basic fear is that several developments — some I described as the modular web, others have to do with net neutrality or commercialization — will lead to a web that no longer is a relatively free, non-discriminating space. It’s a complex topic and I won’t go into any detail here. However, the modular web as I described it could clearly be seen as a catalyst. And that’s very possible.

On the other hand, I am not convinced that there will ever be the one and only solution people use. To some extent we already have evidence to the contrary. As the regular internet is becoming more and more regulated — either by laws or economic dynamics — we today have the darknet. So I have reason to believe that free speech, expression and differing opinions will always find a way to exist online.

That being said, this not a good solution. It’s a plan B or C. We should think about and address this issue as it is never a good idea for societies to marginalize diverging, non-mainstream ideas.

On user adoption

The whole intelligent assistant business creates a very, err, remarkable virtuous cycle. You might be skeptical of AI in general. You might not like to depend too much on technology. Or you might not like the idea of speaking intelligent agents in your living room for (likely very real) security concerns. Still: it will still be really hard to resist this technology once it reached a certain quality. For a simple reason: You must use it in order to keep up with other people using it, for instance from a productivity standpoint. Or, if you want to look at it more favorably: The assistants are to going take the hassle out of the basic tasks of everyday life. Effectively turning it into a decision of principle vs. comfort. A tough one.

Google ahead of its users?

Sundar Pichai says in his keynote:

“We are pushing ourselves really hard so that google is evolving and staying one step ahead of our users”

I believe that is actually how the people at Google think and feel about their business. I also think this is dangerous territory. From a technology/R&D standpoint it’s totally understandable if you want to be ahead of time. But when it comes to putting ideas into practice, you want to get the timing right. And this explicitly includes not being too early for your users.

Daydream / Googles VR approach

I really like Googles approach to VR. After cardboard was pretty successful they now take it a step further and create a premium cardboard if you will. It’s called Daydream and it still revolves around making VR accessible by mobile phone. It a nice way to think about it and — imho — one that is way more suitable for daily use compared to standalone devices.

Google Instant Apps & Modular Web

I didn’t mention it above because it’s not necessary to get the big picture. But Instant Apps is yet another testament to the whole modularization of the web. What it does is really interesting: Devs can start to modularize their apps in a way that they are launched when you click on web links — without you actually having to download the entire app because it only loads the relevant modules. That’s particularly interesting for low-usage apps which you don’t usually download but that offer a better user experience than the mobile website. Shopping away from amazon is a prime example.

What others say around the web

I/O video summary:

Analysis by others

Google Assistant:

Google Home:

The case for devices like Home / Echo by M. G. Siegler:


Future of the web


Machine Learning