Last Tuesday, I tried to plan a research project. I opened Notion for the outline. Google Docs for the draft. Zotero for references. A browser with 14 tabs for sources. Figma for the visual framework. And a sticky note app for the ideas that didn't fit anywhere else.
Six tools. Six windows. Six different mental models. And the thing I was actually trying to do — think through a complex problem — happened in none of them. It happened in the gaps between them, in the cognitive overhead of switching contexts, in the spatial map I was maintaining inside my head because no tool could hold it.
This isn't a productivity problem. It's an architectural one.
Forty years of the same metaphor
In 1984, Apple introduced the Macintosh. With it came the desktop metaphor: files, folders, windows, a trash can. Documents were digital pages. You opened them, edited them, saved them, closed them. The metaphor was borrowed from the physical office — paper on a desk, documents in a cabinet.
It was brilliant. It made computers accessible to millions of people who had never touched a command line. The abstraction worked because it mapped onto something everyone already understood: paper.
But here's the thing. In the 40 years since, everything about computing has changed except the metaphor. Processing power increased a millionfold. Storage went from kilobytes to terabytes. The internet connected every machine on earth. Screens went from monochrome CRTs to 6K retina displays. AI went from science fiction to your text editor.
And yet. You still open a "document." You still save a "file." You still organize things in "folders." You still scroll a "page" — top to bottom, beginning to end, just like paper fed through a printer.
Everything changed. The interface stayed frozen.
The WIMP paradigm (Windows, Icons, Menus, Pointer) was formalized by Merzouga Wilberts at Xerox PARC in 1980. Apple commercialized it in 1984. Microsoft followed in 1985. It has been the dominant interaction model for 40+ years.
The page is a prison
A page has one dimension: vertical. Content flows top to bottom. Ideas come after each other, never beside each other. You can't place a source next to the argument it supports. You can't cluster related concepts visually. You can't see the shape of a project at a glance.
Pages force linearity onto thinking that is fundamentally spatial.
Your brain doesn't process information sequentially. Neuroscience is clear on this. The hippocampus — the brain region responsible for forming new memories — is fundamentally a spatial mapping engine. You remember where things are, not what order they came in. When you recall your apartment, you don't list rooms alphabetically. You walk through the space.
Every tool that forces you into a page is asking your brain to translate spatial, multi-dimensional thinking into a one-dimensional stream of text. That translation has a cost. Cognitive load researchers call it extraneous load — mental effort spent on the format rather than the content.
We've been forcing three-dimensional thinking into a one-dimensional container for four decades. The container was the compromise, not the thinking.
Three things a workspace needs
If we're going to move beyond the page, we need to understand what replaces it. Not a whiteboard — that's too unstructured. Not a database — that's too rigid. Something new. After two years of building Depli and studying how people actually think through complex problems, I believe the answer has three components.
I. Space
The first thing a thinking environment needs is space — not a page, but a canvas. An infinite two-dimensional surface where ideas can be placed, moved, clustered, and connected freely.
This isn't the whiteboard metaphor. Whiteboards are freeform chaos — sticky notes floating in a void with no semantic meaning. What I mean is structured space: a canvas where each element knows what it is. A note module knows it contains text. A source module knows it holds references with metadata. A task module tracks completion. The spatial layout adds a layer of meaning on top of structured content.
Proximity becomes semantic. When you place your research sources near your thesis draft, you're making a statement about their relationship — without writing a single link or tag. Your brain reads this arrangement instantly, the way you read the layout of a room.
Space is not a feature. It's the foundation.
II. Intention
The second component is intention. Current tools are procedure-driven: you tell the computer how to do something (click this menu, select this option, format this text). A thinking environment should be intention-driven: you express what you want to accomplish, and the system composes the response.
"I need to analyze three research papers and find contradictions." That's an intention. In a procedural system, you'd open each PDF, read them, take notes, compare notes, write a summary. In an intention-driven system, you express the goal and the workspace assembles the right configuration: source modules with the papers loaded, a note module for your analysis, an AI assistant ready to help with extraction.
The best interface is not the one with the most features. It's the one that understands what you're trying to do before you've finished explaining it.
— Bret Victor, "The Future of Programming"
III. Adaptation
The third component is adaptation. Your thinking environment should learn from how you work. Not in the creepy, surveillance-capitalism sense — not tracking your behavior to sell ads. In the useful sense: remembering your preferences, learning your patterns, suggesting connections you haven't made yet.
This is where AI becomes genuinely useful — not as an autopilot that replaces your thinking, but as an attentive collaborator that notices things you might miss. "These three notes share a theme you haven't articulated." "This source contradicts your conclusion in the other project." "You usually work on this type of task in the morning."
The key distinction is agency. The system suggests. You decide. Always. The moment AI starts acting without asking, it stops being a thinking tool and becomes an automation tool. Those are fundamentally different things.
Cf. Shneiderman, B. (2022). "Human-Centered AI." Oxford University Press. Chapter 4 argues that augmentation outperforms automation for creative and analytical work.
The shift is already happening
This isn't speculation. The architectural shift from pages to spaces is visible across the industry, even if most people haven't noticed it yet.
Miro moved design collaboration from documents to canvases. Figma made spatial layout the default for interface design. Obsidian introduced a canvas feature alongside its text-based notes. Apple added Freeform — an infinite canvas app — to every device. Even Notion, the cathedral of the page metaphor, added a "canvas" block type.
But most of these tools treat space as an add-on to the page. Canvas is a feature, not the architecture. You still create pages, you still save documents, you still navigate through a tree of nested folders. The spatial layer sits uncomfortably on top of a page-first foundation.
The difference matters — adding a canvas feature to a page-based tool is like adding a sunroof to a submarine. The architecture determines the experience, not the features bolted onto it.
What comes after pages
The answer isn't a better page. It isn't a smarter document. It isn't even a more powerful editor. The answer is a fundamentally different container for thinking — one that starts with space, responds to intention, and adapts to how you work.
That's what we're building with Depli. Not a note-taking app. Not a project management tool. Not an AI wrapper. A cognitive workspace — an environment where your thinking can exist in the shape it naturally takes, rather than being compressed into pages.
The page was a brilliant invention. For 1984. For printers. For a world where digital tools needed to mimic physical ones to be understood.
We understand computers now. We don't need the training wheels anymore. It's time to build tools that match the architecture of thinking itself — spatial, intentional, adaptive, alive.
The page was designed for printers.
The canvas is designed for minds.
Follow the journey
New essays on cognitive tools, spatial thinking, and building Depli.
Join the waitlist →