Why Software Is Disappearing from the World

Co-written with ChatGPT-o1mini

For years, software has seamlessly woven itself into every aspect of our lives. If you could dream it up, chances are there was a venture-backed software startup ready to bring it to life. Businesses that didn't prioritize a digital-first approach found themselves scrambling to navigate the relentless tide of digital transformation. But a few years into the AI boom, everything changed. My workflow transformed dramatically—I’m juggling more specialized tools and paid subscriptions than ever before. The tool stack is evolving at a breakneck pace. What once took hours of manual effort can now be achieved effortlessly by simply asking an AI, watching the desired outcome materialize almost magically before my eyes.

Within the design industry, AI has been both a game-changer and a disruptor. Creating "good enough" designs now just depends on that one great prompt. While mastering effective prompts still requires a deep understanding of one's intent, the learning curve is getting shorter and shorter. Tasks that once relied on specialized professionals—illustrators, graphic designers, UI designers—can now be accomplished through precise prompting. This democratizes design but also challenges traditional roles.

"Good enough is now just a prompt away."

The design space is under significant threat. We're witnessing our own services shift into subscription models. As these subscriptions compete, prices drop. Tools like MidJourney, Canva, and RunwayML empower anyone to create visually appealing content that once required substantial investment. Our social feeds are now flooded with visuals that, just a few years ago, would have been prohibitively expensive or time-consuming to produce.

Handcrafted work is becoming a luxury—not just for visuals but also for writing. Competing with Large Language Models (LLMs) that can generate variants of poems in seconds is increasingly challenging. More than I want to admit, AI outpaces my own creativity. Working with these tools has transformed my role from a craftsman to a curator, gardener, conductor— orchestrating intelligent tools to work in harmony. Understanding analog methods still tremendously helps, such as color theory, lenses, styles, and possessing the vocabulary and eloquence necessary to guide AI effectively. However, newcomers may lack these skills, leading to entirely new outcomes. While foundational knowledge aids in better prompting, visual and written content creation has irrevocably changed. There's now an abundance of content that is more than "good enough" for the tasks at hand.

UI design is undergoing similar transformations. Over the past few years, I've closely followed the evolution of Next.js and libraries like Shadcn. It's become clear that building your own system is a luxury few startups can or should afford. Today's libraries are incredibly powerful, almost complete for any conceivable use case. Starting from scratch is rarely justifiable, especially since we are all accustomed to the styles, affordances, and patterns these libraries have ingrained in us. UI design often reduces then to a paint job, with CSS serving as just another thing to prompt.

"Prototypes speak louder than screens, and screens speak louder than a text file."

The most radical change, however, has been the combination of UI libraries with LLMs. This pairing has fundamentally altered what I as adesigners can achieve. I always believed prototypes speak louder than screens, and screens speak louder than a text file. But today’s tools allow me to go beyond click-through prototypes without realistic data; I can prompt a tool or service directly with intent. While it's still early, experimenting with platforms like Replit or Vercel's V0 suggests that the future of building is heading towards tools that help realize ideas more swiftly and seamlessly.

Visual design has become a commodity, and building tools is increasingly accessible. The TL;DR is that product and visual design have undergone such significant changes in recent years that we may not have fully realized their impact yet.

"Virality demands attention, yet our ability to maintain attention is diminishing."

The consequences are tangible. To attract users today, you must win the "attention warfare." As it becomes easier to build, competition intensifies, driving prices lower and raising the bar for standing out visually and narratively. A strong brand is now just one ingredient in a complex equation. Over the years, I've observed that hype requires a robust position and clear messaging. However, the nature of our feeds—where anger spreads always faster than joy requires more nuanced expression—demands extreme positioning and narratives to go viral. Brands now often resemble influencers who bait audiences into discourse to stay relevant. What marketeers may overlook is that they are under the same threat as any other creative discipline today.

Virality demands attention, yet our ability to maintain attention is diminishing. We scroll through millions of pieces of content, making it increasingly difficult to capture anything that might be meaningful. This is why bold, loud, and extreme content prevails.

An age-old story of the internet is that what can be automated becomes code, and code becomes more autonomous. We are entering the era of AI-empowered profiles. No human can monitor a social feed like an automated system, manage interactions across hundreds of threads, or sustain engagement simultaneously. Consequently, we are seeing more AI-driven profiles that are highly prolific. In that case I’ve come to view bots not as a problem, but as signals of the future—a canary in the coal mine.

As more profiles are powered by AI, our attention becomes increasingly fragmented. Launching such profiles is cheaper and easier, allowing for infinite replication. These AI agents never tire, don’t sleep, and ideally, don’t deviate from their scripts. They represent the dream marketing department—always on mission. However, as their numbers grow, and AI agents begin to interact with each other, generating more content than we can feasibly consume. We are rapidly transitioning from creators to just spectators. To keep up, the best solution so far involves summarization by another LLM, but this merely adds another layer to the already overwhelming flow of content. Traditional advertising and content marketing are under threat from the very tools that make content creation easier right now. Automated social media is, in essence, consuming itself.

"Creating tools and high-quality content has never been easier, but it raises the question: are we truly noticing it?"

Personally, I find myself stepping back from sprawling social feeds, seeking solace in smaller, more intimate communities.

Attention warfare is just one facet of the profound changes underway. Over the years, it has become clear that the internet's future lies beyond 2D surface front-ends. I’m not referring to the virtual reality metaverse, which deserves its own discussion, but rather to the internet as a collection of functions, smart contracts, APIs, and protocols.

"Bots have long been the ultimate power users. They don’t care about beautifully crafted landing pages; they prioritize APIs and instructions for seamless collaboration."

There is extensive discourse about agents happening right now, and while the concept is laden with hype, our current tools are merely early glimpses of what’s possible. The way we interact with the internet is evolving—from manually searching for flight details to envisioning a future where you can simply prompt an agent to book a flight based on your context and preferences.

Agents are envisioned to become the future interfaces. We are shifting away from interacting with services through traditional front-ends to engaging with agents that have access to APIs and contracts. Imagine it as customer service for everything on stereoids. This shift dramatically changes what a product or service even is. The digital software space is transforming permanently. There might no longer be a need to know as a human user the existence of a service because what matters is the ability to articulate your intent into a prompt for an agent to execute.

"What purpose does a crafted brand or UI serve when there is no one of us directly interacting with it?"

Messaging becomes redundant amidst an influx of similar-quality content. Contracts, APIs, and protocols become the backbone, while agents serve as the new front-end. While traditional UIs may still have their place in very niche specific places, the future likely for the common user (e.g. all of us most of the time )resides in mini-apps generated to specific prompts and intents, focusing on single-use one time cases. Traditional UIs, in the long term, are relics of the past.

Where does this leave software overall? It’s hard not to consider that software, as we know it, is disappearing. Agents don’t require UIs that adhere to human design principles. They don't need landing pages; instead, documentation becomes the sole web presence, optimized to answer prompts based on intent effectively. We're already witnessing this shift as creators optimize their documentation to become preferred responses for AI prompts. The future of websites in this case lies in instructions, not storytelling. Traditional marketing, with us as “decision makers,” is also under threat because it's no longer easily discovered in overflowing feeds. How will protocols, contracts, and APIs advertise themselves to each other? What incentive mechanisms will they employ or develop? The reality is that traditional digital design and marketing is gradually coming to an end.

A few years ago, I wrote a thesis for myself called "the internet is not for us," a thesis I believe stronger than ever. We are not designed for this real-time agentic swarm like environment; our human qualities differ from those of an agentic flock of autonomous entities. The laws of physics and biology governing agents simply radically diverge from those we adhere to. We are newton, they are quantum. As the internet becomes more agentic, there's less reason to engage with feeds, scrolls, and spaces. Today's websites and products already feel dated.

Where do we go from here? One certainty is that the design space is shifting from products to agents. This means our interactions will primarily occur between us and agents. There's still much to explore in this realm—we only have so far chat, voice, or new canvases. Most innovation will likely happen when we open the box for new interaction paradigms. Early products like the Rabbit R1 or Humane AIpin although flawed truly hint that interactions are moving away from traditional surfaces, leaving the space wide open for exploration.

As someone whose entire career has been built around digital design, this transformation is both exhilarating and daunting. The notion that websites and UIs may no longer matter is hard to grasp if your salary depends on it. However, there is a burgeoning need for design directly interfacing with agents, where everything requires thoughtful prompts. Visual overload and AI-driven content consumption are radically altering the very concept and understanding for brands, market making, etc. TLDR: The traditional software and digital space is fading.

Every year, I select a few areas to explore further. This year feels different; I've never questioned as much whether my skills will matter in the long term. My conclusion is that we are entering a few uncertain years, but at least short-term there is still a job for me. Although I can't sit still.

In recent years, I've focused on protocols, contracts, blockchains as industrial time systems, reputation systems, collaboration, and market dynamics. This year, my focus will shift to:

There will always be a need for creativity, and agents and AI in collaboration with us should turbocharge it. However, tools and agents need to be designed for emergence, not just execution. Much of our world is built to be gamed rather than played. For me, play is synonymous with emergence. How can we create more space for play rather than competition? How might we design systems that foster emergence as a constant source of product and concept evolution?

AI-backed profiles may be a temporary fad—appealing now but potentially obsolete soon. As agents become more interoperable, the internet becomes more composable through contracts, protocols, and APIs. Swarm-like AI will most likely draw behaviors inspired from biology rather our still crude understanding of 101 economics—how fungi, birds, and fishes move and interact. How value and information is defined or exchanged? How might we design for collaboration, observability, and coexistence in a swarm-like environment?

If we move away from traditional digital design, where do we interact with the internet? If the current internet isn't for us, what is? This involves exploring new form factors, interaction modes, and paradigms. We must question how to interact with an omnipresent yet invisible internet. How might we design for an internet that is everywhere but visible nowhere?

Real-Time

Clocks and time zones have enabled us to standardize and industrialize activities on a global scale. However, time for intelligent agents may operate differently. If an agent has access to all past data and can predict every potential future simultaneously, the concept of time blurs into the present moment. What does it mean for us to operate and interact within a digital environment that resembles a quantum space rather than the linear, consequential space we experience when we disconnect from our screens? How might we navigate a realm that feels fundamentally alien to us? Are blockchains merely our latest attempt to impose control and temporal structure onto this multi-dimensional space? Do agents even care about these constructs?

Traditional Craft

A personal observation: understanding the origins and craftsmanship of traditional tools is immensely powerful for utilizing future technologies. Gaining more proficiency reading and writing code manually enhances my ability to prompt effectively. Photography classes of the past not only improve today my composition skills but also enriched my vocabulary for the prompts I use. I believe that investing in traditional crafts and skills serves as a blueprint for my future. Whether the internet fades away, leading us to cherish these skills more, or they become the foundation for what's next, they remain invaluable. In my vision, the future specialist is a well-rounded, real-world craftsperson who can seamlessly translate their expertise into digital formats natively.

Software and digital design are disappearing in their traditional forms. While the exact timeline is uncertain, the transformation is undeniably massive. Agentic actions based on intention and prompts are revolutionizing our interaction with the internet. This shift means that the surfaces where interactions occur are evolving. To support our ability to state intentions and translate them into prompts, we need to design for emergence with better feedback loops. Agents that collaborate are likely to resemble natural systems in an ever changing quantum state. Embracing traditional crafts and deeply reflecting on them will enable us to better curate and orchestrate our prompts and intentions.

The past has taught me that working with the best early teams through WE3.co has been the best way to observe, discover, and shape the future. This year, I plan to start building more on my own.

If you want to explore any of these concepts more openly and creatively, please reach out.