A version of this essay was originally published on Medium.

Many of us have been using the dev builds of iOS 11 and MacOS High Sierra this week. I wanted to share some thoughts on what I think are some of the important advances.

I have attended WWDC for many years, sometimes as a partner (working on Office), sometimes as a competitor (working on Windows), and sometimes just as an interested developer (grad school). There are always a range of emotions coming out of the event. In this era of massively global Apple, every event is galactic in scale, yet it is super important to keep in mind that this is still their developer conference.

My view is that iOS is the healthiest developer ecosystem right now. For sure, a big part is because the most revenue is accruing directly to developers, but just as important is the reality that the most valued targets for advertising and commerce are also using Apple devices. Together, these create a vibrant and lively community of developers actively working on updates and new apps. An even more important reason for me is the unprecedented scale of customers and ability for Apple to deliver new software APIs to developers that will make it to hundreds of millions of devices in short order.

While the ecosystem might be the most robust right now, we also know how quickly the landscape can change. That is why the importance of how Apple evolves software that takes advantage of their market position is so critical. I believe this WWDC had some incredibly interesting developments in this regard.

I won’t cover everything, of course, but I wanted to point out what I think are the most interesting innovations or opportunities for developers engaged in the platform. Think of this as the trip report I would bring back for the team, after using the software for a week.

Think of this as a conference report, not a review or coverage of the entire event — what rose to some level of strategic importance. Opinions are mine, and I of course have no insider or unannounced roadmap knowledge of Apple (or any other platform company mentioned).

The tl;dr version of this post would be:

• iOS 11: This is where there are significant advances in the kinds of apps that can be built for mobile devices. This comes from two areas. First, the changes to core user-interaction models for iOS bring what many believe are important features for “productivity” to the iPad (now iPad Pro). Second, new frameworks, but particularly for augmented reality, are moving well beyond incremental and addressing needs in building today’s apps. Both ARKit and Core ML are likely more interesting than I think many believe —democratizing two key aspects of computer science that will clearly differentiate platforms going forward. Privacy remains an extremely core theme to how the product is evolving and how Apple thinks about the way apps are implemented.
• macOS High Sierra: Apple probably surprised people with some of the “features” in High Sierra, but what should impress people more are the architecture and infrastructure advances. Apple has cleverly engineered work for iOS to “trickle down” to macOS, which picks up the APFS file system and video pipeline work. These are a big deal for iOS, but frankly an even bigger deal for macO,S where it is much more difficult to deliver these reliably, given the openness of the platform. There are a good share of advances in High Sierra that benefit developers as users (developers also includes creative professionals more broadly).
• Hardware Products: Delivering on “Pro” with the iPad Pro and iMac Pro is significant. The iMac answers the need for a high-end workstation, and does so with arguably the most powerful device available, yet in a sleek all-in-one form factor. The iPad Pro finally (almost?!) puts the iPad in a position to be a laptop for the masses, especially those who grew up only on phones. There’s a new holiday gift for everyone with the new home speaker HomePod, but in introducing this product, Apple can take this category (if it is one) in a different direction, like it did with home media players when Apple TV was introduced.

macOS High Sierra

From the first words of the macOS section, I was intrigued. I was intrigued not by the boldness, but by the expectation setting of what would follow  —  a focus on perfecting the platform. I was personally reminded the challenges of getting fans excited while also trying to achieve some level of excitement when unveiling a product release viewed as refinement.

Apple doesn’t do anything by accident, but I do think they significantly undersold not just the results but their own execution capabilities in discussing High Sierra. Part of this comes from the always difficult task of articulating innovation in an OS to a non-technical or business audience when an OS can (and should) fade to the background and let apps do the work. Part of this comes from the reality that the innovations were, as Craig Federighi said, “deep technologies for the future”.

Apple File System: AFPS was not the first thing to be discussed, but it is the first thing worth noting, simply because it puts the notion of “perfecting” in context. APFS is an entirely new file system enabling such features as clones, snapshots, encryption, 64-bit limits on file counts and sizes, crash protection, crazy performance for large file operations and more. It is really hard for me to overstate the incredible nature of delivering a file system.

I’ve lived through all the Apple migrations and all the DOS/Windows migrations, and not only is this among the most feature-rich releases, it is actually running right now on my Mac (and iPhone) after an in-place upgrade. I seriously sat there watching the install process thinking, “This is going to take like a day to finish, and it will probably fail and roll back in the middle or something.” After about 30 minutes, the whole thing was complete. The amount of amazing engineering that went into both this creation and deployment of APFS is mind-blowing. And that it was done on phones, watches and PCs is nothing short of spectacular, and except for maybe the transition from FAT to FAT32, I can’t recall anything even close to this. There are a ton of features under the covers that will surface in use of Apple devices, but mostly it will just make everything better seamlessly.

Safari: Most everyone in SV and most tech enthusiasts use Chrome on the Mac. There are some old memories of a slower, less compatible browser with “fewer features” (read, extensions) that keep that audience off Safari. I switched and ran into zero problems. The further refinement of features such as intelligent tracking protection and overall what was referred to as “serene browsing” are more worthwhile to me (and probably most people) than extensions in Chrome.

There’s some jealousy in Apple’s ability to pull off anti-tracking measures, as “Do Not Track” was a huge failure for me when I was working on Windows and IE. Apple’s anti-tracking is really about accountability — if you don’t visit a site, then it is likely the site should not be tracking you. It is a first step. Apple continues not pushing the browser as a platform, but pushing the browser as an application. This becomes very interesting, because the innovation in the browser platform might be past peak for now, and thus the developers out there looking to have differentiated (versus required) web experiences is minimal (perhaps except for Google). When a platform is relatively static, it becomes more important to compete on the quality of an app experience than on the differentiated use of new and incremental platform features.

Video pipeline: Improving the video pipeline with H.265, HEVC, is another one of those things that is really hard to demo and take note of today but over time, this is a critical innovation. For video in and out, hardware support is critical, and Apple has done a great job here. Expect other ecosystems to have variations in support across devices. The challenge, of course, is that this will become a bit of a standards battle between H.265 and Google’s open, royalty-free VP9. We’ve all seen this movie before. Because YouTube commands such a large amount of video usage, this will not get resolved simply or quickly, and one should expect some “wedge” applications or uses.

Thunderbolt Dev Kit enclosure: In a world where devs that need (really need) discrete graphics for rendering or significantly driving external monitors, this is both a welcome and important advance. This is the kind of thing that brings both excitement and some frustration to the dev community. It is exciting because this solves a real problem with the Mac line, albeit with a really big dongle. It is potentially frustrating because it isn’t clear (yet) just how broad and cheap the support for external graphics this will be  — how many graphics cards, how much futzing, etc. The eGPU space has seen lots of attention, and at the same time the forums are filled with tips and tricks to get things working. Generally, as much as there is demand from a community for this, the ability to drive heterogenous graphics cards or live-swap cards from one kernel has always been a bit of a stunt. And. of course. the implications of arbitrary drivers to core system security are significant.

Other OS features: There were a slew of features in the OS at the “app layer,” as well, many or most of which are shared with iOS. As we’ve come to expect from a modern OS, there’s yet another revamping of notifications and control center stuff. Frankly, this is driven quite a bit by synergy with iOS, but because of the different expressions, one mostly had a “moved my cheese” feeling on a laptop. This all falls clearly under the “refined” moniker. Historically, Apple has been so minimal at moving around OS things, much to the disappointment of fans; interestingly, things seem to be moving more each release these days.

Mac Hardware

iMac Pro: The announcement of the iMac Pro is significant, and also shows the Apple approach as we’ve come to expect. This is one beast of a PC — really a workstation —and for $5000 it should be (maxes out at 18 cores, 128GB, 4TB SSD, 10Gbs Ethernet, 4x Thunderbolt, 2 x 5K display capable, etc).

That said, one would be hard-pressed to build a competitive device on Linux or Windows for significantly less, though expect many apples-to-oranges comparisons (sorry for the pun) that allow for fans of each platform to claim price/performance superiority. When you consider the new file system, new graphics APIs, new video pipeline and more, this is likely going to be the most powerful workstation available that at the same time provides the “out of box” quality and reliability that one would expect from Apple. Oh, and it has an SD card slot, which honestly is nice for most hobbyists, but mostly puzzling.

Some might believe that by packaging the device in a “sealed” and/or all-in-one form factor, it will be perennially behind the curve. In practice, this market is served by the slow-to-change “workstations” from major vendors and by DiY gamers. The former are incredibly well-served by this PC, and the latter will continue to build their own, often challenged by compatibility and variations in driver/systems support.

iOS 11

I’ve been running iOS 11 all week on both phone and tablet. Of course, it is not ready at all for daily use (not unexpected), and you can’t really see the benefits of the platform just yet, and many of the usage features aren’t really there yet. That said, the key takeaway for me is that there are some significant “market expanding” elements of the release that are deeper than just adding more frameworks.

From a developer perspective, the capabilities in this release of iOS, combined with the broadening device ecosystem (Watch and HomePod), are furthering the gap between iOS and Android. This is introducing important choices for developers. I continue to believe that “winners” in categories will integrate and exploit native platforms, and viewing every advance through a cross-platform lens is an innovation disadvantage. It is always important to remember that few customers own multiple platforms, so being consistent across platforms solves your problem but not customer problems.

ARKit: One measure of the potential for a new framework is how fast new applications show up that show off the capabilities. In the case of the new ARKit (augmented reality), demos showed up before the morning sessions the next day (The Verge even catalogs some, and @a16z’s @kylebrussell has this repo). The first thoughts about AR tend to gravitate to gaming, especially with that being such a big category.

My view is that ARKit will become closely connected with business scenarios across commerce, training, assistance, trials, advertising and more. It’s not hard to imagine a brand creating some “product placement” experience that goes viral. While “glasses” AR platforms are busy trying to recreate a phone and phone-software platform that either fits in glasses or can be easily tethered, Apple already has a device and platform at massive scale. Frankly, it is not hard to imagine a peripheral that takes advantage of this and brings hands-free. Or maybe not, since this capability will be so rich and accessible. With this release of ARKit, Apple will start the flywheel of API->Apps->Feedback->Features that is the hallmark of new and differentiated capabilities.

ARKit demo made day of show (@CodyBrown)

Core ML: Much continues to be said about Apple missing out on or even structurally incapable of taking advantage of machine learning (ML). Is Apple inherently a client/device company? Is Apple’s lack of a scale services platform the problem? Does privacy hold Apple back? My view has always been that there is a significant amount of cross-customer learning/data that is required for the broadest and most horizontal features  —  it is impossible to implement maps and the full range of routes without using a lot of data. That said, Apple’s aim is not to enable more broad horizontal services with the newly introduced Core ML, but to enable ISVs to build on the data they have/collect to deliver useful features and to then offload some of those uses to the device.

On the one hand, this is simply a way of using the computation on the device to deliver faster response; on the other hand, it is a way of providing features that might minimize data collection on device and thus improve privacy. The main parts of Core ML are vision tracking, natural language and models. Vision permits tracking, face and object tracking, text detection and so on  —  note the connection to ARKit as this type of layering is something Apple does so well. NL supports a bunch of features that greatly simplify what was previously rule-based and algorithmic work. Models provide access for using either existing models for basic classification (for images, for example) or to run your own models. Again, these features further the gap between device OS platforms (not just features, but how apps are structured) while significantly advancing the state of the art. Many will debate the merits of some of these features on the client, models in particular, but I believe that there are far more uses for scenario specific and vertical learning that can and should be done on the client, without shuttling data to services.

Tablet experience: The fun part of a developer conference is always the part where end-user features are shown, since everyone can relate, including the non-developers watching. The advances in launching, switching and working in general geared toward the iPad user are significant  —  so significant that there’s a good chance we are at a turning point where many more people will use (or just admit to using) their iPads for core productivity work.

Yeah, I’ve been saying this for years  —  that ARM-based mobile OS, with new apps geared to a new interaction model, will become dominant. I didn’t expect that to be just a pocket-sized device, but based on hours of usage, that is clearly the case (and at least partially responsible for iPad sales curves). What I believe Apple has cleverly done is introduce features such as “windowing,” drag-and-drop and app switching that will cause the industry to take note of the improved productivity potential while at the same time not forcing a “desktop” model on “everyone.” By and large, these features are likely to fall to power users, but that is often how markets tilt. The new Files app (which is very early) will prove to be a game-changer, and so clearly ups the “power” of the device as many core productivity scenarios are about juggling multiple files in some workflow. For the vast majority of people that define productivity as “Office” scenarios of notetaking, slides, lists, basic models, communicating (iOS was already the preferred mail platform by volume), and so on, the iPad, with its security, reliability, robustness, performance and also connection to phone (continuity, Messages, etc.) make for an extremely productive experience. Developers take note, as iPad-specific apps will become increasingly important in productivity categories.

One of many demos of iOS 11 files app (this one by Jake Chasan)

Messages: For developers, Messages has made apps more visible and easier to get to (and in the beta really difficult to remove without crashing). Messages is almost certainly the most-used app in U.S. (and potentially European) markets, but in those markets, the idea of apps in messages has not caught on. There’s an interesting opportunity global companies have, which is when adding features for one market (say China) can turn those seemingly “odd” features into global use cases. In this case, it is not clear if the “rest of the world” will follow WeChat to a messaging-centric world, or if there will be a whole generational upheaval to the mobile platform, the way we saw the uniqueness of Japan’s mobile infrastructure upended by smartphones rather than the adoption of that approach. Building an app for Messages is an odd investment right now, though of course the way Apple scales the platform makes this incrementally small, relative to building an app (as it does with Watch and Tablet support).

SiriKit: SiriKit was introduced last year, and for the most part it is the ability to add the rough equivalent of Alexa Skills to Siri. Perhaps because of the overall reputation for Siri, or maybe because of the excitement around Alexa, this hasn’t seen the noticeable takeup one might expect. It could also be that voice control is still too much of a gimmick, especially when your phone is attached to you. iOS 11 expands SiriKit with more domains (built-in apps of notes and lists) and support for payments, QR codes and ride booking.

More UI changes (camera, photos, CarPlay, Ink):  iOS 11 also includes many improvements to the basic use and integrated apps. Most of these do not substantially enhance or differentiate the developer opportunity, but are an important part of advancing the overall ecosystem. You can read or experience these elsewhere.

However, the one major change worth noting is the new screenshot workflow, which is seriously game-changing. Not only does it address common capture issues (getting the wrong shot the first time, but still saving to camera roll, or always needing to crop out the carrier/time), but it adds easy markup, annotation, sharing without saving and more. Closely related is the ability to capture video of sessions, as well (but not video of sessions using the camera or capture!). In addition to random Ink, you can add text, a signature, shapes/callouts or add a magnifier. Crazy cool!

For Pencil fans, the Ink work is very nice, and shows Apple’s fundamental approach which is that Ink is like an acetate layer on top of everything or a region to fill, but not a keyboard replacement. There are many new features, some of which have existed in Windows (or Windows or iOS apps) for quite some time, such as shape recognition, that will prove valuable to some. The CarPlay “do not disturb” feature is exactly the type of feature that Apple can and should do, that few others would try, and so it is worth high praise for this work. Most platforms would think of CarPlay DND as too parental or even taking away freedom, but it is absolutely the right thing to do.

iOS hardware

Like many iPad users, I’ve been surprised and disappointed at the pace of iPad hardware changes (as I’ve tweeted about often, iPad is my primary device in terms of hours of usage, and where I do writing, financials and slides, in addition to longer sessions of reading, social, etc.).

The software improvements in iOS 11 already make using a 9.7-inch iPad Pro better. But nothing makes new software seem better than new hardware.

The nature of information or knowledge work (selling, marketing, product management, financials, planning, executing and more) continue to undergo changes. These changes are never happy overnight, and so many look to the pace of change to insist that the change isn’t happening. At the same time, so much of the public dialog around work in technology is dominated by engineers and designers who can and should be using Mac or Windows PCs (and will continue to do so). That whole world is 50 million to 100 million people worldwide, and isn’t likely to change for some time (until the scenarios completely change, or those tools exist on an iPad).

The new iPad Pro looks is exciting, and having had a chance to see one, I am certain that a broader set of people will tilt toward iPad-only (assuming that we can still fly on planes with iPads!). The typical knowledge worker will move to the iPad — or some other device with ARM, mobile OS and those attributes — sooner rather than later. The iPad gives us a big screen to accompany our phone-sized screen.

The new iPad Pro with a slightly bigger screen is great. The new keyboard continues to disappoint (me). It is still a keyboard in the middle. It is too bulky for occasional use, and too constrained for all-day use. The origami unfolding is awkward, the lack of function keys limiting, and the stability poor. I have used the Logitech clamshell, and long for a much slimmer and better-designed clamshell container, which I know Apple (or anyone) is capable. There’s a new Pro case that turns the iPad into a Surface (like RT) with full-sized keys and a kickstand. It is still heavier and more awkward than it needs to be. That’s the one I’ll be using until a better clamshell case comes along (or Apple just does a clamshell iPad).

One more thing

HomePod was introduced at WWDC, even though there was not a direct developer message beyond SiriKit.

Music is a blind spot for me, so there are many better-qualified people to discuss this scenario. On the other hand, home control is a huge area for me. It is worth noting that in the real world, music is essentially infinitely more important than home control.

I’m aware of all those who use Alex/Echo and now Google Assistant for home automation tasks. As aware as I am, I am more skeptical that these will advance beyond demonstrations for some time. I’m especially dubious that a stationary speaker will serve as a broad entry into controlling home devices. The fact that the devices from others look to be adding to or relying on screens seems to scream “doesn’t quite work.” The idea of a stationary but smart screen in the home just seems weird when everyone has a screen with them or really close by all the time. And much like AR glasses makers rushing to make a full mobile OS/platform for glasses, speaker makers recrafting an existing OS to create yet another app platform seems tricky, especially if it is an interactive platform.

Home control and automation really doesn’t work at scale, except for a few silo scenarios now (Ring doorbell is a great example of something that works broadly). Something needs to replace one-off apps, but it isn’t clear to me that the replacement is me memorizing a vocabulary for each device and speaking to a stationary device.

That said, one needs to look carefully at not just the what (a really nicely done premium entertainment device) but the how. Apple has set expectations, and is likely to deliver a “just works” product that looks to deliver on the core scenario of music even better than competing products. That’s what Apple did with portable music and, later, video. It was after that initial entry that Apple chose to expand the developer opportunity and platform capabilities of the product. Perhaps with Apple TV, Apple failed to capitalize on that approach. Perhaps with Watch, Apple tried to do too much too soon, and did not deliver on the just works aspect. HomePod seems much closer to how iPod and iPhone came to be, and for that reason alone, I am optimistic.

Apple released a massive amount of software this past week. Most of it works, too! MacOS is ready for most any developer to use full-time in my experience; iOS 11 is less so.

There’s a tendency to see a cloud of analysis around Apple events looking at things from a perspective of Wall Street or how iPhone unit sales will be impacted, which I think can be short-sighted. What Apple showed at WWDC expands computing in new ways  —  new capabilities for mobile phones that can be used by a scaled ecosystem, new hardware combined with software capabilities that can (finally) change how typical business users accomplish productivity tasks, and hardware and software for consumers that bring a new take on the execution of existing by still-immature categories.

Steven Sinofsky is a board partner at Andreessen Horowitz, an adviser at Box Inc. and an independent investor and advisor in Silicon Valley companies. Follow him @stevesi.

This article originally appeared on Recode.net.

What do you get when you mix a choice selection of the leaders of the most important companies, a deeply engaged tech-savvy audience, and the best interviewers around? You get Code Conference. This thirteenth (13!) installment of the unique conference proved as timely, revealing and interesting as ever.

While the conference has been through a few changes of ownership, some name changes and different venues (and seat configurations), the constants of Walt Mossberg and Kara Swisher — along with a deep roster of Recode writers supporting them — make for the place to hear from (and on video, see) how technology and business trends are created, used or handled by the important companies of our time. On a personal note, I’ve been lucky enough to attend every one of these — you can find me on the far side, taking notes and listening attentively.

By any name, machine learning, artificial intelligence, deep simulation — or, as IBM’s CEO insisted, cognitive computing — were front and center for nearly every speaker.

Some years are remembered as launch years (like the iPod, or even Windows or Office releases!), and others are marked more by debates over disruption (such as net neutrality or music distribution). Rarely do we get to experience a year when the breadth of speakers collectively express both optimism and execution plans for a technology shift as we saw this year. By any name, machine learning, artificial intelligence, deep simulation — or, as IBM’s CEO insisted, cognitive computing — were front and center for nearly every speaker.

Over the course of the previous 12 conferences, there have been themes that everyone might have touched on, but I don’t recall a case where there was such uniform aggressiveness at staking a claim on the technology. Across the board, speakers went to great lengths to talk about how their customers are going to benefit from the use of intelligence technologies (let’s call this “AI”) in products and services.

Why is this not simply a waypoint on the way through the hype cycle? The short answer is because we, as customers, are already “using” intelligence every day on our smart phones. It is worth pausing to acknowledge that somehow over the course of the past year, AI has gone from passing reference through implementation to daily use.

Today’s use of AI is not hype, but reality.

Across the board

The most fascinating aspect of listening to the speakers respond to questions about how AI will or does play a role in their respective enterprises is how the technology spans devices, strategies and business models. While each of us might be familiar with a specific example, looking across the speakers paints a picture of an incredibly rapid and deep technology diffusion.

Amazon CEO Jeff Bezos spoke at great length about the role of AI in the company’s breakout product, Echo. In many ways, Echo has come to symbolize the true potential of multiple technologies across voice commands, agents and machine learning, all packaged up in super-simple consumer devices. In addition, Bezos outlined how Alexa — the technology underpinnings of Echo — is both a customizable platform and an embeddable technology. Developers can build new “skills” for Alexa and contribute “learning” to offer new capabilities (and as Mossberg noted, owners receive weekly email detailing the latest skills added). Makers can embed Alexa technology in their own devices — one example mentioned was an alarm clock — and turn a mundane technology into another AI-enabled endpoint.

Ford CEO Mark Fields worked to convince the Code crowd that while Ford’s intrinsic strength in AI or cloud might not be there, they are actively partnering (with Pivotal, for example) and bringing in-house the skills needed to make sure Ford vehicles participate in this technology wave. While Fields did not articulate a deep vision of AI, he used examples around maintenance and navigation to illustrate the company’s commitment.

In an approach similar to Ford, Cisco CEO Chuck Robbins talked about committing his company’s thousands of software engineers to AI, and incorporating it into Ford’s existing product lines. He spoke of using AI to bring better management and understanding to modern networks that will contain millions of end points of all sorts.

While spending a good deal of time differentiating his company from Amazon, eBay CEO Devin Wenig talked in detail about eBay’s amazing work to improve the shopping experience, and to eliminate fraud by using AI. Wenig characterized the use of AI as the way the company will deliver a highly personalized eBay — one that is curated and relevant. Fraud, he claimed (in a follow-up interview by Lauren Goode, from Recode’s Vox partner The Verge), has become a “meaningless” number through the use of technology.

That AI is so mainstream today that dozens of CEOs can articulate company execution plans using the technology is directly attributable to four important changes in the technology landscape.

IBM CEO Ginni Rometty traced IBM’s longstanding efforts in AI, and outlined what amounts to a “bet the company” investment in the technology. She discussed scenarios from health care to education, from business IT to third-party developers, and from cloud to on-premises as ways IBM is working to contribute to and support the use of AI. Most interesting were the efforts IBM is making to provide open source or freely available solutions, which Rometty said ran counter to IBM’s history, but were essential to see the company moving forward. While many of us think of IBM’s Big Blue or “Jeopardy” building up to Watson, IBM’s history with AI goes to the very start of the field, where the famed Watson Labs worked to pioneer the earliest ideas in translation, language processing, speech and handwriting recognition, and more. Even with so many challenges, one has to be impressed by Rometty’s outlook on the technology and the depth of IBM’s engagement.

Perhaps the most frequent use of AI any of us experience is through Facebook on our mobile phones, and together onstage, Facebook COO Sheryl Sandberg and CTO Mike Schroepfer outlined numerous efforts in AI across the site, advertisers and, of course, in virtual reality. The depth of both product features and applied research going on in AI at Facebook is not as widely appreciated, I believe. For example, a few weeks ago, Facebook contributed a dozen papers to the major AI conference ICLR. As mentioned below, the role each of us play as people (not “user,” according to both speakers) is a huge contributor to Facebook’s ability to deliver AI-based features like photo tagging that we find so valuable.

Fresh off the heels of the I/O Conference, Google CEO Sundar Pichai took us through the history of AI at Google. There is clearly no other company that has the depth of efforts and the broad use of AI in products over time. Without a doubt, AI has always defined Google, and it is only in the past year or so that this has become broadly understood. Photos, inbox, search, advertising, Assistant, self-driving cars — and the list continues —were all examples used by Pichai to illustrate Google’s ongoing commitment to AI. If AI is itself a platform, Google is most certainly the most invested and best positioned to be the leader in such a platform.

Why this year?

The Code Conference has certainly seen disruptive shifts before: Digital media, mobile devices and smartphones, to name a few. AI is proving to be a different kind of shift — not one that some are resisting or concerned about disrupting their legacy business, but one that everyone is embracing and running toward.

Down the road, the interesting question will not be which companies used AI, but which companies made the most of AI in novel ways.

Down the road, the interesting question will not be which companies used AI, but which companies made the most of AI in novel ways. There is a massive amount of inventing left to do in the field, because even with the rapid rise of the technology, things are still very much in their infancy. For example, Skype Translator was previewed by Microsoft two years ago at the Code Conference, and is built using the most modern AI techniques of deep learning.

That AI is so mainstream today that dozens of CEOs can articulate company execution plans using the technology is directly attributable to four important changes in the technology landscape. I think it’s worth reflecting on how these came together to make it possible to see such a rapid move from lab to deployed feature in such a high-tech endeavor.

Raw compute power for models: While we love to talk about Moore’s law as enabling so much, when it comes to AI, it is Moore’s law applied to parallel architectures, not Intel’s scalar ones. The application of even more transistors to graphics processing units (GPU) has been a key enabler of AI technology. Cloud architecture plays an incredibly important part in this because companies do not need to build out their own GPU data centers in order to tap into the power of AI training models, but can tap into on-demand scale.

Massive data capacity for training: Everyone has come to learn that more data is the only way to train AI models, and you can never have too much data (but it is easy to have too little). It is only recently that cloud architectures have become “routine” and “economical” at both retaining and accessing the quantities of data required for training. Facebook provides the easiest-to-visualize view of this as it trains recognition of people on the more than 300 million photos a day (or half a petabyte). IBM’s examples using radiology offer another view into just how important the evolution of storage is to AI.

In the span of a short time, AI has made a leap, and has likely skipped over the trough of disillusionment. I am not saying that lightly.

Incredible availability of labeled data: As important as data is to training, data without “labels” isn’t very helpful. This is where our role using technology plays an important part, as well as the openness of the internet. It isn’t just that we upload photos to Facebook, but that we tag people we know, and in doing so, train the image-recognition engine. It isn’t enough for eBay to say it wants to offer a customized store, but that we are signed in and purchasing items to inform the customization engine. This is such an advance over the way we used to think of click streams or guessing if someone is a return visitor.

In addition to all of this, sensors in our phones offer motion and location data, enhancing everything we are doing. Clearly and obviously, there are privacy and security questions with all of this, but at the same time, never before have there been such personal benefits to each of us as we use services. The availability of data goes beyond that which I personally generate (and label), and includes data sets and APIs that are now available simply because of openness and cloud-based solutions (for example, economic and demographic data from governments) that can be incorporated as part of training models.

Open implementations of technology underpinnings: The most fascinating aspect of the rise of AI solutions has been that so much of the core technology has been developed in the open (often by the research arms of companies) or is at least contributed in an open way relatively early in the evolution of the technology. Google’s TensorFlow, Facebook’s Torch, IBM’s SystemML and UC Berkeley’s CAFFE, along with technologies for data (such as Spark) are all openly available platform elements. This is most certainly following the same pattern as HTML/HTTP, which means the economics will be elsewhere in the system (in the data, training and models, of course).

What’s next?

In the span of a short time, AI has made a leap, and has likely skipped over the trough of disillusionment. I am not saying that lightly.

While there is no doubt that some will be disappointed in what transpires over the next couple of years, there is also no doubt that such skepticism will be communicated through a vast number of writing and communication tools all benefitting from AI. Nothing ever does everything everyone wants as soon as everyone would like.

It is clear, without qualification, that AI is a mainstream technology among the technology leadership companies, and in the near future will be an ingredient of most every leading product and service.

Steven Sinofsky serves on boards of several Andreessen Horowitz investments, and is an investor and adviser to startups and an adviser at Box Inc. Reach him @stevesi.

This article originally appeared on Recode.net.

Recently, frequent Re/code contributor Steven Sinofsky was having a Twitter exchange with CNBC on-air editor Jon Fortt about “speaking up” in the workplace. Sinofsky was about to post this essay on the subject when discussion emerged yesterday over Microsoft CEO Satya Nadella’s remarks at the Grace Hopper Celebration of Women in Computing, being held in Arizona.

Hardly a week goes by where we fail to see further evidence of the challenges women face in the workplace, especially in our technology industry. I have been reflecting on my own efforts as a manager, and which actions have had positive results. There are actions that colleagues — particularly managers who are men — could take to foster a better environment to hire, support and promote a diverse team, especially with respect to women.

We continue to see more women in leadership and management roles in technology companies, but we know that the majority of those roles are inhabited by men today. While all managers can do more to be more inclusive for all people, this post has suggestions for actions that male managers can easily take today to create an environment more supportive of women on the team.

The following is a set of “everyday” things you can do, starting immediately. They are easy. They almost certainly require a behavior change. They will make a difference — I know they did for me.

These actions, however, are not a complete solution to the broader challenge, which is important to emphasize. There is amazing work going on, and much more and new amazing work is needed.

Know the women on your team. Seems obvious, right? When you have a startup, you know every hire. When a team is very large, it becomes impossible. But most of us are in the middle of that spectrum. It is important that, as a manager, you make an extra effort to know the women on your team. What are they working on, and are they set up for maximum potential success? Do they have the right support/mentoring opportunities? Are the work group and management environments functioning as they need to? Why or why not?

It is too easy for you to let the internal network or organization dictate whom you interact with. If more women are in part of the team that you don’t see as much, or are at a different spot in the organization that you don’t interact with often, then you won’t get to meet with them by “accident,” so you have to make an extra effort. In most tech organizations, there are fewer women than men (and fewer in leadership), so you want to be deliberate in making sure you put in the extra effort to meet them.

Measure performance on achieving goals, not on heroic work. Again, this concept probably sounds obvious. It is always the case that you want people to set goals and achieve them. One pattern I observed in the early days of Microsoft (when it was decidedly male) was how we would tend to reward the behavior pattern of “sign up for more than you can do, scramble at the end to get some of everything done, and finish up during final testing, long after things should have been done.” This doesn’t work well, even though we continued to reinforce this pattern. Why? Is this an issue for women on the team? While women can just as easily demonstrate this goal/achievement pattern, they more frequently demonstrate the pattern of signing up for the amount of work they can do, and doing it. Crazy, I know. To some this looks like sandbagging. To others, it looks simply like “promise and deliver.” I learned this from the first woman development manager (from the ’80s) that I worked with. That’s why “promise and deliver” was always how we ran our teams.

Invite women to lead team meetings, not just to do the demos or act in support. When you have a big team meeting or company event, you need to make sure those presenting represent the full diversity of the team. When you plan on this, don’t force the issue, and don’t just default to the org chart. Find a natural way to be representative and inclusive and not have obvious token representation. It isn’t so difficult. Why? A really bad thing you can do is have the women on the team play a support or secondary role to the main event or speaker. In tech, “doing the demo” is a role like that. Not participating in Q&A, for example, is the wrong signal. Doing just the introduction sends the wrong message. I made a mistake like this once in 1994 and really never lived it down. (I asked a woman on the team to draw the door prize at a user group meeting — good grief, what was I thinking?)

Ask the opinion of women in a meeting, no matter where they are sitting. Good managers always solicit opinions from other people in the room. If the room is filled with people who have no problem raising their voice or interrupting others, then people who don’t exhibit those behaviors won’t be heard. Keep in mind that the people in the room demonstrating those traits are also probably not listening all that well. So when you are running a meeting, make it a point to proactively seek out the opinions of those who were just listening. Why? Again, any person can exhibit any behavior, but far more frequently, the men in the room are doing the talking and interrupting, and the women are not. So if you want to hear the opinion of the women in the room, make sure you stop to do so.

Hire, promote and mentor women. This sounds so obvious. Your role as a manager is to lead at these things. How often have you asked recruiting to deliberately focus on hiring women for the team? Have you given them feedback on candidate slates or interview loops? How often have you personally declined the opportunity to mentor a man so you can allocate that same time to mentor a woman (everyone has finite time — how are you spending your time)? When meeting with other managers about promotions or new assignments, have you been the one to be proactive in support of women, or to hold off until there is more diversity in potential candidates? There’s risk in all of these that you will “lose” good candidates or frustrate some internally, but you do have to start somewhere. Are you? Why? Because this is the heart of the issue, and no matter what anyone says, you can do these things now.

Talk to everyone in the hallway and lunchroom, not just to people like you. Andy Grove was among the pioneers of managing by walking the hallways. The interesting thing is, who do you talk to when you walk the hallways? You should be deliberate. It is natural for everyone to talk to people they have some connection to or similarity to beyond the obvious (same team). Whether that is college, kids, hobbies, movies, TV, sports, or even which part of the food line you visit, these all contribute to the natural flow of people. Break your own patterns. Why? Do this because it might not come naturally to you, and because it is easy to do, and because it matters. There’s a natural tendency for like to seek out like. As a manager, you should do more than that. If all your conversations in the hallway are about “man sports” and “man hobbies” with other men, and people hear those, you’ll only reinforce stereotypes. Instead, make it a point to talk to women on the team in these spontaneous conversations.

Obtain feedback deliberately. All managers want and ask for feedback, but how and when you ask for feedback is just as important. Ask for feedback in 1:1 hallway conversations and direct email, for example, not just with an “any feedback?” throwaway line right at the end of a meeting or event. Why? A non-zero set of people are reluctant to share feedback in front of a group or right away. Some might believe that their feedback is unique to their situation, or that offering the feedback might single them out in a negative way. Similarly, as above, some prefer to reflect and synthesize before giving what might be visceral or of-the-moment feedback.

These are just a few things from my own experience that I’ve worked on over the years. I wasn’t always great at these, and certainly the organizations that I managed and was accountable for were far from what they needed to be. I can say that I was deliberate about these behaviors and actions. It is behaviors like these that are necessary to address the inequalities in the workplace, though not sufficient.

Former Microsoft executive Steven Sinofsky is a board partner at Andreessen Horowitz, an adviser at Box Inc., and an executive in residence at Harvard Business School. Follow him @stevesi.

This article originally appeared on Recode.net.