If you sat down on a Sunday evening and tried to draw, on a single sheet of paper, everything you worked on last week — the tasks you completed, the conversations you had, the documents you reviewed, the decisions you made, the projects each of those belonged to, the people they involved — you would not draw a list. You would draw a web.
The shape of working life is a graph. Lists are a flattening, calendars are a flattening, even most project management tools are a flattening. Each of them is useful at the moment of capture and limited at the moment of recall, because the structure that gives an artefact meaning is its connection to the artefacts around it, and most software immediately throws those connections away.
The Me Graph is the part of the Oestler Notebook that doesn't throw them away. It is a fully three-dimensional, explorable visualisation of every record in your personal workspace, rendered in your browser, with edges drawn wherever a real relationship exists in the data.
What's In It
The graph collects records from across the Notebook and the project surfaces a single user touches: to-do lists and the tasks within them, kanban boards and their list memberships, Janus AI chat threads, distilled memory fragments, daily summaries, projects, document register entries, drafts, annexures.
Each becomes a node. Each node has a colour and a size derived from its type. Each is positioned by a force-directed simulation that pushes everything apart while pulling connected things together — exactly the algorithm that produces the familiar organic shapes of network diagrams, except here it runs in three axes rather than two.
The edges are not decorative. They mirror real foreign-key relationships in the underlying data. A task is linked to its list. A list is linked to its kanban board. A memory is linked to the chat or task that produced it. A consolidated long-term memory is linked to each of the fragments that were merged into it. A draft is linked to the document it derives from. The graph is the schema, made visual.
You Are At the Centre
At the literal centre of the graph is your own profile node. It is bigger than everything else, marked with your avatar if you've uploaded one, and connected to every top-level record you own. From there the graph radiates outwards: lists, then tasks, then memories, then chats, then the side branches into kanban boards and project documents.
This isn't only a stylistic choice. It is a useful organising principle for a workspace that grows over time. The closer a record sits to the centre, the more directly it belongs to you; the further out, the more it sits in shared territory. Pan the camera around the centre and the structure of your engagement with each part of the platform becomes visible at a glance — the dense cluster of tasks belonging to one project, the long thin chain of memories from a particular reflective conversation, the isolated nodes that haven't yet been linked to anything else.
The Me Graph isn't trying to tell you what to do next. It is trying to show you the shape of what you've already done, in a form your eye can process faster than any list ever could.
How It Renders
The graph is drawn with hardware-accelerated WebGL. We use instanced geometry — a single mesh, drawn many thousands of times in one GPU call, with per-instance colour and position attributes — which is how the graph remains interactive even when it represents tens of thousands of nodes.
The physics simulation runs in a Web Worker, off the main thread, so the canvas stays smooth even while the layout is settling. Every few ticks the worker posts an updated array of node positions back to the main thread, which transfers them straight into the GPU buffer with no copy. This sounds like a technical detail but is the entire reason the graph works on a laptop in a browser tab: keeping the simulation off the rendering thread is what separates a usable visualisation from a slideshow.
The camera controls are the same vocabulary you would expect from any 3D viewer — orbit, pan, zoom — with a keyboard toggle to switch between an orbital camera (which pivots around a focus point, ideal for inspecting a region) and a free-look camera (which moves through the graph as if you were walking through it). On mobile, touch gestures map to the same operations, with the layout simplified and the node count capped to keep the experience snappy on lower-powered devices.
Click a Node, Open the Record
The graph is not a museum exhibit. Every node is interactive. Click a task and you see the full task, with its history, its comments, and its current status. Click a memory and you see the fragment, the source it was extracted from, and a button to mute it from future Janus conversations. Click a project and you jump to the project's main view. Click a list and the graph isolates only the nodes connected to that list, fading the rest, so you can see the cluster of work it represents.
This is the feature we use most ourselves. The standard list views in the Notebook are excellent for sorting, filtering, and pure execution. The graph is for the moment when you want to recall context — to find the conversation that produced a task three weeks ago, or the cluster of memories that came out of a particular project's commissioning phase, or the orphan note you wrote on a Friday afternoon and haven't connected to anything since.
Why a Graph and Not a Search Box
A reasonable question, when looking at a feature like this, is: why not just give the user a better search? The honest answer is that search and graph solve different problems.
Search is the right tool when you know roughly what you're looking for. You type a phrase, you get a ranked list, you find the thing. It is a precision instrument for retrieval.
The graph is the right tool when you know there's something there but don't know what to call it. You see a cluster you don't immediately recognise. You see a node connected to three projects when you expected one. You see two memories sitting close together that you'd never consciously connected before. The graph is a recall instrument, and an exploration instrument, and — quietly — an audit instrument, because the visual shape of a workspace makes its inconsistencies apparent in a way that a list view never quite does.
Both are valuable. The Notebook offers both. The Me Graph is the one you reach for when search gives you the answer to the wrong question.
Privacy and Boundary
The Me Graph is, as the name implies, personal. It is rendered for one user, from data that user owns. Shared lists and shared kanban boards appear in the graph if they are shared to you, but the graph never aggregates other users' personal records into yours. Memories are particularly carefully scoped: only the memories Janus has extracted from your own conversations and tasks are included. There is no team-wide Me Graph and there is not going to be, because the value of the visualisation comes from its specificity to the person looking at it.
If your organisation needs a graph view across an entire project — including assets, documents, and other users' shared records — that is a different feature, called the project graph, and it lives inside the Spatial Hub and the project's main views. The Me Graph is the personal complement: your slice, your shape, your nodes.
What It Looks Like in Practice
Open the Notebook, switch to the Me Graph tab, and the canvas fills the workspace with the cloud of your own data. The first few seconds are settling: the simulation starts in a tight ball and breathes outward as the forces resolve. After about three or four seconds the layout stabilises, and what is left is a shape that looks startlingly like the structure of your week.
Hover a node, see its label. Click a node, see its record. Drag the camera around the centre, see your projects bloom out from your profile. Press I to isolate a cluster, F to fly to the selected node, R to reset the view. There are no other controls, because there is nothing else the graph asks of you. It is one feature, doing one thing, doing it well.
The Me Graph is available in every user's Notebook today. We think it is one of the most quietly useful things we've shipped this year, and the kind of feature that takes a few weeks of regular use before its value becomes obvious. We'd encourage you to try it for that long.