We engineers love things that work. We find a pattern, use it successfully, refine it and use it again. We build reusable components, establish standards and share solutions so that the next person does not have to start from scratch. That instinct has helped us build an extraordinary digital world.
But somewhere along the way, repeatability can become a limit. A familiar pattern stops being one possible answer and becomes the answer. We reach for it before we have fully considered the problem, because we know how to build it, we know how to maintain it and people already know how to use it.
I have been thinking about what happens when we challenge that instinct. What if the way we have become accustomed to building websites is also preventing us from seeing what the web could become?
I call this idea the Possibility Web.
We built what we knew how to build
The web did not arrive fully formed. It developed over time, shaped by the tools, skills and limitations of the people building it. Pages, menus, categories, search boxes, forms and checkout flows became common because they solved real problems. They gave structure to an enormous and growing digital world.
People learned those structures too. They became comfortable finding the menu in a familiar place, searching for a product and adding it to a cart. Businesses learned to expect certain layouts, and engineers developed reliable ways to deliver them. There is value in that familiarity.
But imagine someone walking into a physical shop and being handed a map of the company’s stockroom. They are told to find the right aisle, identify the right shelf, compare the boxes themselves and return to the front once they have worked out what they need. They might learn to do it quickly. They might even become very good at it. But we would still have to ask whether the experience was designed around the person or around the way the stockroom was organised.
Much of the web asks us to do something similar. We adapt our questions to fit search boxes, translate our needs into filters and move between pages to assemble an answer. We are so used to doing it that we rarely notice the work involved.
Something built around yesterday’s limitations and familiarity may be part of today’s problem. It might still work, but “it works” should be the beginning of the conversation, not the end.
We should be setting the trends
There have always been people willing to look at something familiar and ask whether it has to be that way. Think of a fashion designer sending something unexpected down a runway. It might look impractical or strange at first, but the designer is exploring an idea beyond what people are used to wearing. Some of those ideas are eventually refined into things that become part of everyday life.
Technology has its own examples. What would mobile phones look like today if Steve Jobs and the team at Apple had accepted that a phone needed a physical keypad? They challenged a familiar part of the device and gave more of its surface to the screen. That decision helped open the door to ways of using a phone that went far beyond making calls and sending messages.
As engineers, we often adopt ideas like these once they have been proven. The unusual becomes familiar, the familiar becomes a standard, and eventually we are building it without questioning where it came from. What if we spent more time being the people who question the standard in the first place?
For a long time, there was a practical reason to stay close to what we knew. An idea that was too far outside the box could take months or years to build and test. If it failed, the cost in time and effort was substantial. Established patterns gave teams a way to deliver something useful without taking that risk every time.
AI has changed that equation. Ideas that once took months to explore can now be built and tested in hours or days. Some can reach people and grow far more quickly than before. There is still work involved in making software dependable and ready to scale, but the cost of finding out whether an idea has potential has changed significantly.
Yet we are often still thinking inside the same box. We use new tools to build the same pages, the same journeys and the same experiences, only faster. We have gained the freedom to try ideas we might once have dismissed, but we have not always given ourselves permission to use it.
We have an incredible view of the digital world. We understand how information moves, how systems connect and how experiences are constructed. Now we have more opportunity to put unusual ideas in front of people, learn from them and see where they lead. Instead of only asking, “How do we build the next version of what already exists?”, we can ask: “Why does this need to exist in this form at all?”
Start with what people are trying to do
Consider someone who wants to create a comfortable home office. They might have a small room, a limited budget and no clear idea which desk or chair would work best. Their goal is not to visit four product categories. Their goal is to make a space where they can work comfortably.
Or consider someone planning a trip. They may know they want a few quiet days somewhere beautiful, but not know where to go, what it will cost or how much travel they are willing to tolerate. The moment they begin looking, they are asked to make a series of decisions: destination, dates, flights, accommodation. Those questions matter, but they may not be the questions the person is ready to answer yet.
In both cases, the person begins with an intent. They have an idea of the outcome they want, even if the details are still unclear. The way they explore may help them discover those details.
That is where I think we should begin to reimagine the web. Rather than making people fit their intent into the structures we have already built, what could an experience look like if it helped them explore that intent? What might they discover if they could see different directions, understand the trade-offs and change their minds as they learn?
I am not suggesting that every shop or travel website should look a particular way. I am suggesting that we stop treating the current way as the starting point for every idea.
Make room for possibilities
We often design as if the user arrives knowing exactly what they want. Sometimes they do. But often they have a feeling, a problem or a direction, and they need help understanding the choices in front of them.
Think about buying a home. A person might begin by wanting a larger house, then realise through exploring that what they really need is a shorter commute and a room to work in. The process of looking changes their understanding of the problem. The best answer is not necessarily hiding behind the first search they made.
The Possibility Web is an idea for an internet that makes more room for that kind of discovery. It would help people explore what could be possible, understand what matters to them and follow a thought as it develops. It would not assume that the first question is the final question.
That also means being honest about trade-offs. A different choice may cost more, take longer or mean giving something else up. Those consequences are part of the decision. Helping someone see them clearly can be more useful than simply presenting an answer.
There should also be room for surprise. An experience that only shows people what it expects them to want can make their world smaller. Sometimes the most valuable possibility is the one a person would never have thought to ask for.
AI gives us a reason to rethink the human experience
A lot of the conversation today is about AI agents doing things on our behalf. An agent can take an objective, ask questions, look for information and work towards an answer. That is exciting, and it will change how many tasks are done.
But while we explore what agents can do, we should also explore what people can do with better ways to interact with the digital world. A person may want an agent to investigate something, then want to examine the possibilities themselves. They may want to question an assumption, explore a different direction or make a decision based on something the agent cannot fully understand.
This is not a choice between human exploration and agentic systems. The interesting space may be in how they work together. An agent can help with the work of finding and organising information, while the person remains free to steer, question and discover.
That makes trust essential. If an experience is helping shape a decision, people should be able to understand the assumptions that matter. They should know when information is uncertain and have room to disagree with what is suggested. An experience should help people see more clearly, not simply become more persuasive.
Code is easier. Thinking is still ours.
For a long time, the cost of building limited how far we could explore. A team might have had an unconventional idea but chosen a familiar approach because it was faster, safer or easier to explain. That was often a sensible decision.
Today, the distance between an idea and something we can try is shrinking. We can explore directions that might previously have remained a sketch on a whiteboard. We can test an unusual interaction, learn from it and change course with less effort than before.
That does not mean every experiment should become a product. It does mean we have fewer reasons to dismiss an idea before seeing what it might teach us.
This is where I think our role as engineers becomes even more interesting. Producing code is becoming easier to access, but knowing what to question, what to explore and what matters to a person remains deeply important. We bring knowledge of what is possible in the digital world. We should use that knowledge to imagine new experiences, not only to reproduce the old ones more efficiently.
Of course, we still need dependable systems. We still need to care about security, accessibility and the people who have to maintain what we build. Thinking outside the box does not mean throwing those responsibilities away. It means applying our engineering judgement to a wider range of ideas.
We need people who do not think like engineers
I have been watching people use AI to build software without an engineering background. Call them vibe coders if you like. Some of the things they try are things I would never have thought of. My first reaction is sometimes, “That is not how we do it.” But then I look at what they are trying to achieve and ask myself: “Why not?”
I have seen ideas that feel completely out there, yet make sense once you see them in action. These builders are not carrying the same assumptions I am. They have not spent years learning which patterns are considered sensible, which ideas are usually rejected and how an experienced engineer is expected to approach a problem. Sometimes that means they miss risks an engineer would catch. Sometimes it means they see possibilities an engineer would overlook.
I think that is a perspective every engineering team that wants to evolve should have: someone who can build and experiment, but who is not bound by the usual engineering way of thinking. A non-engineer using AI to explore ideas could show a team what it has stopped questioning. The engineers can bring the knowledge needed to make an idea dependable and maintainable; the other person can help them see an idea they might never have considered.
Of course, that only works if engineers are willing to listen. It is easy to dismiss an idea because the person presenting it cannot explain how they would scale it, secure it or maintain it. Those are important questions, and engineers should ask them. But they are not always the first questions to ask when someone is trying to show us a different possibility. First, we can try to understand what they have seen that we have missed.
Learning is part of the job
One of the things I have always felt about software engineering is how much it asks us to keep learning. Other engineering disciplines evolve too, and I do not want to underestimate their complexity. But software moves at a pace where the tools, practices and possibilities around us can change dramatically during a career. What we knew yesterday is valuable, but it is rarely enough on its own.
That is why I find our response to AI interesting. Many engineers have spent their careers adapting to new languages, platforms and ways of building, yet some are reluctant to engage with this change. Perhaps part of the discomfort is that our experience does not automatically make us the most experienced person in the room at using these new tools. Someone outside engineering may have spent more time experimenting with them and discovered approaches we have not considered.
We do not have to accept every claim about AI, and we should not abandon the judgement we have developed. But refusing to learn from a tool—or from someone using it differently—feels at odds with the curiosity and adaptability that software engineering has always demanded of us.
Will engineers listen to a vibe coder? That is their choice. I think the teams that do will give themselves a better chance of seeing what comes next.
Let’s ask why
I am building a demo around this idea, starting with e-commerce because shopping gives us a familiar experience to question. But the Possibility Web is bigger than shopping. It is a way of asking how the internet might work if we started with human intent and allowed ourselves to rethink the structures around it.
I do not have a finished blueprint. I do not know which ideas will work, which will feel too strange or which might lead somewhere entirely different. That is part of what makes it worth exploring.
I would like us to get more comfortable asking why things are the way they are. Why does a person have to follow this path? Why do we divide an experience into these steps? Are we doing it because it helps, or because it is familiar to us as engineers?
Some answers will lead us back to what we already know. Others might lead somewhere unexpected. Let’s try the ideas that feel a little too far out there, learn from them and see what they make possible.
We have the knowledge of the digital world at our fingertips. We have tools that make exploration easier than ever. Let’s use them to question what comes next, rather than only rebuilding what came before.