Hardware as a commodity

Why buy a computer when you can run a browser on a phone? Your computer is worried about your local file permissions while all your life is hacked somewhere on the Internet. The internet hardware islands has overlived its usefulness.

In Reflyx hardware computing devices are critical but not central to the platform. For many reasons expressed in Rethinking software, we believe that software must be freed from its hardware jail and use abstract computing capacity when needed. Since the emergence of the Internet, software has been deployed and executed on multiple hardware devices. Each device still being a discrete computing entity and software having to be predesigned and deployed using the right protocol to seamlessly communicate and execute on these computing nodes. This is the cloud we have today. Someone else's computer(s).

Why do you need to buy you own device in 2024?

Most of the computing is done in the cloud anyway. You need a device for the camera, the screen on which you consume content from social networks running in the cloud. You send message and perform a few voice calls, everything done mostly outside your device itself. Not surprising that innovation in phones' software is slowing down. If not for AI features and the support of new form factors or hardware devices, software changes would be mostly cosmetics or internal to help reduce maintenance costs for the manufacturer. Even today, mobile network providers bundle the phone with their network subscription for an extra fee. They need you to have a more powerful device than you can afford to pay for their newest network technology, so they help you get a better phone. In the hardware-centric world, this is the closest thing to the future of computing we foresee. We're closing in on commoditized computing devices.

The world's your docking station

When you use electricity, you don't have to generate your own, you just plug in the wall socket and you get it. What if we could do the same for computing capacity? The world made you believe that you need your device and a fast network to be happy, but what if what you need is computing capacity? What if hardware devices were hidden in your house walls or provided by a provider's cloud wherever you go? What if you could just transport/wear a small device and use the world around you as a big docking station, with various form factors and specialized computing like gaming, VR, transports, shopping, banking, healthcare and more. You wouldn't need to buy your own powerful device. Just a way to hook into a powerful computing grid and securely get access to your content, wherever you are, without the need to sync you devices or manage local files.

To achieve this level of hardware abstraction, we need to drastically change our perception of computing, software and the way manufacturers profits from computing devices. Take a look at Nvidia, with their Geforce Now offering. They rent their hardware instead of selling it. They give you access to top of the line GPUs for a fraction of the cost and let you stream the result directly to your own device. You don't need a powerful device or even a very fast network to experience top level gaming. This is a achieved using proprietary software applications on both sides and limited to their ecosystem. In Reflyx, we bring this model to all industries and on steroids.

Computing capacity in Reflyx

Reflyx, being a actor centric operating system, completely decouples the computing capacities from the actual resource model. From the actor perspective, everything is done within the confine of its own domain (potentially through a cockpit for humans). The domain is device agnostic, virtually managed by the platform and made accessible on the current device without the user intervention. You trigger a mutation, it will reserve the proper computing capacity and execute within the target domain, through a pre-established secure subscription. Computing capacity is thus everywhere, exposed by nodes, which can take the form of specialized computing devices like appliances, VR rooms, cars, mobile phones, laptop computers, a grocery cart, a bank teller, anything.

Actor centric computing

The diagram below provides a high-level representation of one actor, with multiple domains, exploiting the Reflyx platform through various UXStations.

When powered on, devices will instantiate a node, which will connect to the Reflyx platform. The device has computing capacity inherited from the underlying hardware and host kernel (Linux at maturity). This capacity has to be provided to the grid, which will be done by subscribing the grid to the device publication (remember, devices are actors too). A device may have multiple publications offering mutliple usage profiles but devices publish their capacity to the grid just like any other actors on the platform. The grid will thus have a collection of computing capacities with various attributes and specializations. Attributes like physical location, cost, availability, proximity, capacity will be handled by the grid to select the right computing capacity for a specific actor request.

In the solutions section, you'll find many different sample designs of the reflyx computing model applied to specific industries or problem. You'll see the relative simplicity of the overall solution compared to today's fragmented computing islands.

In this diagram, we extend the previous architecture to show two actors and how they can subscribe to each others' publications. The location of the subscribed content is irrelevant in this model, it is logically mounted inside each actor's root domain, through an explicit subscription detailing how this information can be consumed or even modified (through exposed mutations). As explained in other sections, the information is never duplicated, always retrieved from a single source of truth. If, for performance reason, you decide to cache it, it will be at your own expense, using computing capacity (storage), which will be reactively updated when the source changes.

New Computing Business Models

Hardware will not disappear with the new Reflyx computing model. Hardware products will evolve and manufacturers will figure out new way to profit from their computing technology. Just like Nvidia renting their GPUs in their cloud gaming platform, manufacturers will rent their computing capacity, in a way that is quite similar to mobile network operators today. This is the next logical evolution of our computing landscape as more and more devices are rented from mobile network operators, paid monthly and returned when the network subscription is terminated. Reflyx is just extending this model to offer so much more flexibility to both manufacturers and network providers, but also, to a whole new breed of vendors, offering AI services like ChatGPT or very specialized processing capacity as we'll discuss below.

Portable Hardware Identity Device

The phone as a computing device form factor has been around for the past 20 years with growing capacity. But, in the new computing paradigm, phones don't offer much added value when combined with infinite computing grid and embedded UX stations everywhere. You don't always need a screen, speakers or even a camera when you can get these for free or for a minimal fee around you. Instead of the phone concept, we think that Reflyx users will move to small identity devices, either carried or biologically injected or connected, that will unlock their subscriptions to surrounding computing devices in the home, car, work or anywhere in the city. Proximity connection to a UX station like a screen, a camera or a VR room will connect this identity content and let the user access its data or perform transactions. You want to text or call someone, just ask your UX stations, through voice recognition to connect you.

These devices could have their own UX station like our current phones, market and UX station evolution will determine which form factor will be more successful. One things for sure, interaction with surrounding UX stations in your day to day life will be very fluent, simple and completely secure. Compared to all so-called smart WIFI devices in your house, each one pushing unsecured usage data to their manufacturers or to a central vendor (a.k.a Google, Apple or Microsoft), the new computing model will combine all your information in your private cryptographically protected space, unlocked by your portable identity device.

Computing Capacity Providers

In reflyx, all actors, not just humans, subscribe to computing capacity to perform their operations or even to exist. Until humans have the capacity to interface directly in their brain, they need two types of computing capacity: the data processing capacity and the UX station to support their need to interact with their domain. The only thing humans will have is their identity (usually as a hardware device) to they need to connect to a UX station through a subscription and get access to their domain and attached computing capacity.

Instead of generic computers, Reflyx will open the way to various specialized UX devices with a variety of technology and form factors. Here's a short list, but many more could and will emerge with Reflyx adoption:

• interactive walls
• grocery carts
• shopping stations
• VR rooms
• laptop UX
• phone UX
• car dashboards UX

So these UX stations will be installed in public area or acquired with your house or car. For the nostalgics of those who wants to execute good old applications, they will still be able to buy laptop-like appliances with verry simple hardware connected directly to the computing grid. So from the UX perspective, this will be mostly use-case driven. You visit a clothing shop, you connect to their UX station using your identity (you don't have to share personal details) and you subscribe to their catalog, interacting with it in your own domain. You can involve AI actors in your buying experience, your purchase history, you bank account, anything your already have in your root domain will be accessible to interact with the catalog, always securely in your own space, not in the store or shop.

We can foresee vendors (like mobile network operators) rent small UX devices (like phone?) acting as both a minimal UX and a hardware identity device, connected with a subscription to a variety of computing capacity and addon services (like AI, media processing, specialized data processing and much more). Instead of installing a WIFI router in your home, they might install a Reflyx node, with computing power provided to the platform, with some retained for the subscriber usage. Sky is the limit! Moving from low-level network providers to high-level computing provider will open so many business cases for these players, that they will push for Reflyx adoption from the get go.

Computing Farm Operators

Another business model, analog to the current cloud providers, would be organizations providing thousands of nodes, with generic and specialized computing capacity, publishing their capacity to the grid. This model is pretty straightforward and will surely emerge and be quite useful in powering fluid software solutions. One innovation that will arise is per-transaction computing pricing, which will let multiple vendors bid on transactions to lower computing costs and use a sane market approach to regulate the computing capacity of the platform.

Appliances subscriptions

So called "smart" appliances are a computing category by itself. These were the initial specialized computing devices connect to their manufacturer or through the Internet. With the current computing paradigm, the use of smart enabled devices was pretty limited. But with a modern computing grid like Reflyx, these devices can leverage subscriptions and contribute to their owner's domain instead of sending raw data to their manufacturers. One business model, the appliance rental, capitalized by the manufacturer or by external investors, will surely be a big hit as it helps manufacturers create products that will last, instead of forcing renewing at the expense of planet resources.

Creating connected objects, exposing very simple publications and computing capacity will be cost-effective in Reflyx. Device manufacturers will not have to embed too many computing components as long as they provide a WIFI or other connectivity capacity. Onboarding of new devices will also be simplified as all these devices will publish to their owner's domain, secure and no need to send arbitrary data to their manufacturer.

Going Further

You can dig deeper into Reflyx commoditized computing model by studying some of the solutions we described in the solution section. While all of them use the computing grid in some for or another, the more interesting ones from a hardware computing perspective are the following:

Reflyx Car Computing

In this solution, we design a software system applicable to the full lifecycle of a car, including manufacturing, operation (driving & repair) and various other perspectives like insurance, spare parts, financing, warranty, fleet management and much more. Such a system would be nearly impossible to create using the current computing paradigm as it is very hard (if not impossible) to create a reality snapshot that would define exactly everything that will happen during the lifecycle of a car. This Reflyx solution is not trivial neither, it involves the transformation of a whole ecosystem to succeed, but it shows a clear way how cars could leverage fluid software, commoditized and specialied grid computing and be used as a UX docking station when driving.

We also explore how AI will be integrated in this solution, either for self driving or for more trivial improvements like traffic avoidance, energy economy, trip planning and maintenance management.

You can find all the details in the Car Computing solution.

Reflyx Laptop Showcase

Part of our vision is to transform operating systems and thus, we need to explore how personal computing devices will work in the new reality. The Reflyx Native Computer solution explores how such a computer would be designed and how you could still, even in a commoditized computing world, own your own laptop or desktop computer.

The high-level concept is that you get exclusive access to the computer computing capacity, just like today. You get access to the computer UX station, similar to what you have right now. While this is a simpler solution than the car above, it still involve multiple actors and a very different mindset. Such computer could be rented (see above) and could expose multiple computing publications, depending on what you want to do with the computer. The computer itself could be virtually attached to grid computing resources to increase its capacity or power specialized workloads (like AI or gaming) without having to physically install the hardware.