Generative Surfaces

Analogs help us imagine a parallel universe to the one that exists. One logical leap away from a new world. Rather than a horse, we imagine a car. Rather than a car, a plane. Similar, but different.

Over the past hundred years, we've almost achieved instant gratification for the physical world. Slowly we've graduated upwards from small, local merchants to large, "a lot of things" stores like Walmart to eventually the "(not really) everything store" via Amazon. These were all logical analogs as the world scaled up.

Amazon achieved what we previously thought impossible by orchestrating our physical needs in real-time and dealing with all the complex logistics to move atoms around the world in mere days.

But it begs the question: what would the Amazon of the AI world look like?

The Amazon of AI will be any digital OR physical experience you want... but it may not be Amazon who creates it.

You Are Here

We've tirelessly built so much over the short time since we invented software. We started with rigidly configured hardware that needed expensive changes to be reconfigured. We finally figured out how to make hardware transient and malleable at an electrical and firmware level, opening the door to software as we know it.

When we achieved the "software-defined" era of hardware, a shift started to happen. Each of us could personalize our experience, with hardware being an afterthought as long as it just worked. In fact, people almost stopped talking about the hardware altogether and instead focused on their experience of using the software. The story changed to concrete benefits versus abstract desires.

Shortly after, software began to feel just as rigidly configured as the early days of hardware. You could only do what it was originally configured to do and little else. If it didn't exactly work for your use case, you either dealt with it as a "bad alternative" or you moved on in search of something else that had slightly fewer tradeoffs. Rarely do people build exactly what they need, because building today requires an enormous effort across an increasingly long tail of needs (both in the present and in the future, for maintenance).

Now we're up against a wall, a dam of sorts that temporarily holds back everyone's frustration that software can't personalize itself even more to their needs. As we developed Viv, we saw this with Siri where it did very little to evolve itself around our worlds and instead forced us into its own way of doing things. Everyone eventually chalked it up to a failed attempt and "personal assistant" became an ugly word.

This is obviously a privileged position compared to early hardware days, but as humans we're hard-wired to continuously optimize our own reward functions in life. Whereas other animals seem to accept compromise more willingly, for humans it seems to never be good enough.

Where You Want To Be

"Be where you are. Otherwise you will miss most of your life."

But, where do you want to be? Increasingly, this means opting out of mundane jobs, mundane companies, mundane (& static) locations, and now: mundane tools.

We yearn to explore and discover. We push to evolve ourselves as much as possible before we take the great leap into the abyss and darkness of death, life's greatest change agent. This means we optimize our time, shedding weight that holds us back from advancing.

We yearn to create.

Sometimes we create for others, but we first strive to create something that we're individually proud of. Often before we have the bravery to share, we need pride in our creations.

In the past, being proud of a creation meant spending months, years or even decades in pursuit of a mastery that gave you some perceived superpower to create in ways that others would admire.

Over time, these initial barriers towards creation are being lowered which enables a much higher ceiling for all the possibilities and variability of creation because more people will be able to create in ways that resonate with them and others.

But what will it take to enable anyone to create... anything?

The Generative Surface

Every day we use phones and laptops to interact with the world. Our entry point to the internet is often a web browser or an app that someone else created, on a device that can only really operate a specific set of apps and functions.

It's effectively hard-coded, moving only at the pace of other people's ingenuity and resourcefulness.

This won't be true for much longer.

Soon, we'll all be able to create at the speed of thought — and this will extend beyond just the digital world to then touch the physical world so we can truly shift our realities to better match our internal desires.

Generative Devices

It's inevitable that Apple, Google, Microsoft and others will be forced to collapse their entire devices into a generative app surface. The future also doesn't look like 3rd party app bloat; that era has been deprecated.

Instead, you'll have a phone or laptop that learns with you and adapts itself completely to your needs, constantly breaking down barriers and building new utilities to keep you happy, engaged and even entertained.

The big change will be that the device itself won't host much of the software or compute. Instead, the device will be a portal to a generative AI factory: an AI operating system.

Phones and computers will effectively be a simple internet terminal. There's no need to have local compute and storage anymore, beyond orchestrating simple hardware functionality like location, camera, microphone, motion, etc.

Software Factories

Humans have so far failed to reliably outsource software in the same way that manufacturing & hardware builds are routinely outsourced.

AI will change this and the underlying teams. The software factories of the future are being built right now, and the workers won't (often) be human.

Apple has bet big on local device capabilities, and it might come back to bite them now that compute is the largest currency in AI. Ideally, software will be infinitely elastic and the best way to do that is in the cloud (where Apple has historically struggled).

This opens the door for anyone to create an AI-native phone that does very little other than accessing a generative software factory in the cloud.

These factories will enable anyone to summon software with a few utterances of their voice. When the software inevitably falls short or needs evolving, the factories will modify and personalize the software for the end user specifically — as single-use, potentially throwaway software.

And it won't stop at software.

Physical Factories

Software is a bridge meant to be destroyed and rebuilt. We pave the road, we tire of it, we pave a new road and so on. In this way, software enables us to kill off workflows or experiences that stopped working. No scar tissue or physical clean up, just a shuffling of transistors.

These software bridges will help us build workflows into physical factories, laying the groundwork for puzzle pieces that interconnect from one to another until we eventually can manifest anything we desire into not just digital experiences but physical ones.

Even phones and computers will fall into that category when they all get decomposed into recipes that allow anyone to configure their own hardware that simply accesses a generative AI factory in the cloud. Hard tech won't be so hard anymore.

Legos

Everything becomes abstractions and then everything becomes legos.

Amazon's shopping website abstracts away factories and airplanes, Amazon AWS abstracts away server racks and running cable, Google search abstracts away web scrapers and humans writing content to seek your attention.

But, those abstractions aren't enough. Today's abstractions can't be used by anyone to create anything.

They can't be easily composed together into infinitely variable experiences and physical products. So before we can build that, we need to decompose what already exists and make it effortless to rebuild it, as well as experiences very similar or entirely different to it.

Soon, every app will have an IDE built into it; it just won't look like an IDE — and it won't be exclusively for software.

The world is about to enter the era of primitives — everything becomes legos and then legos become everything.

Single-User Experiences: For you, By you

Your toddler, mother and grandfather are all about to be able to build & share just-in-time experiences; I’m not sure enough people realize the gravity of that.

When the user is also the builder, what use do they have for software teams?

This will cause an industry-wide restructuring & resizing of teams, forcing a lot of people back out of software while collapsing “middleware” roles supporting it (product managers, marketing, engineering, design, user researchers, executives, support, you name it).

Users will be able to build what they want, when they want, and at minimal cost.

Future moats will be more about building the primitives and focusing on translating user needs into built product, since switching costs will be very low in the future.

Distribution won’t matter.

A lot of founders talk about distribution being a big competitive advantage, but a lot of people get this wrong about the future.

The new moat will be solving people's unique problems instantly with just-in-time experiences.

When you can build your own software or physical product in minutes, there's nothing to distribute.

And what happens if it breaks? It’ll likely be self-healing, but it'll also enable people to hot-swap experiences and components from a collective pool of other software and physical components.

It will be recipes, except for experiences.

Bon appétit.

-- Rob