Building Notion for the post-AGI era

The same pages, written together

We're building Post Notion: a workspace where you and your agents write the same pages, and your work is to consume, steer, and share what gets made, on any device, with nothing carried by hand.

That sentence is the whole product. The agent is the author. The page is a means. Your judgment is the point.

Alan Kay predicted the shape of this in 1972. Most of a personal computer's use, he wrote, would be "reflexive communication of the owner with himself through this personal medium, much as paper and notebooks are currently used." Notion proved him right for a generation. People used it to think. The next medium keeps the reflexive loop and adds a second party to it: the loop now runs between you and your agents, and it runs all day.

You read it, hear it, or watch it. You answer with a decision. The decision goes straight back to the agent. The page keeps itself current. The bar holds. That cycle is the heartbeat of the thing.

Your job is to consume, steer, and share

When a thousand docs can be generated on demand, the doc gets cheap. The judgment that says this one is right stays scarce.

So the home screen shouldn't open on empty storage waiting to be filled. It should open on what needs you: the drafts your agents made, the calls that are yours to make, the work sitting at the line waiting for a yes or a no. The question changes from what you want to write to what your agents need you to decide.

The durable unit of work is the decision: what got approved, by whom, and against what standard. A doc is a snapshot. A decision is a permanent fact about your company. Slack loses the decision when the thread scrolls. Git keeps the what and rarely the why. Most of the time the reasoning lives in one person's head and walks out the door when they leave.

A company should hold its decisions the way it holds its code: versioned, attributed, searchable, and permanent. Get that loop right, where the agent keeps the page true and the human curates by reviewing the change, and the rest follows.

Context is the moat that compounds

Models get cheaper and stronger every month. Raw generation is on its way to free. So the advantage lives somewhere else: in something a model can't ship with.

Friedrich Hayek named that something in 1945, writing about the use of knowledge in society. The knowledge that decides outcomes, he argued, is local and situated. It is "the knowledge of the particular circumstances of time and place." It "never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess." Swap his central planner for a general model with no access to your company, and the argument lands the same. The model has the general rules. It doesn't have your particulars.

Hayek went further, in a line that reads like a brief against the omniscient model: "To assume all the knowledge to be given to a single mind in the same manner in which we assume it to be given to us as the explaining economists is to assume the problem away." No single mind gets all the facts. They are scattered across your team, your approvals, your rejections, the thing you learned last Tuesday and never wrote down.

A workspace can capture that. As generation gets commoditized, the scarce input becomes trusted, organization-specific context: what your team has approved, rejected, and learned. Stronger models widen this gap, because a sharper agent does more with good context, so the team that has captured the most pulls further ahead every quarter. Models are rented by everyone. Context is owned by you. That is the one moat that compounds.

The org brain, and why the work never ends

Douglas Engelbart saw the endgame in 1962. The value would migrate from any single piece of work to the structure connecting it, and one good capability would reshape the whole. "A direct new innovation in one particular capability can have far-reaching effects throughout the rest of your capability hierarchy," he wrote. Capture the decisions and the context behind them, and you have changed how every future decision gets made.

That is what Post Notion becomes over time. The living brain of a company: the one place where agent work arrives, gets judged, and compounds into context that makes the next round sharper. The decisions carry forward, and so does the reasoning behind them, inherited by whoever comes next. Knowledge that is held this way keeps compounding. David Deutsch calls this the beginning of infinity: knowledge well made has reach, generalizing far past the problem it was built for, and the supply of good explanations never runs out.

Some people will hear "the machines do the producing" and brace for the human role to shrink. It expands. Automation takes the grind: "The 'perspiration' phase can always be automated," Deutsch writes. The judgment stays, because "problems are soluble," and solving one only reveals the next. He is blunt about why that is good news: "an unproblematic state is a state without creative thought. Its other name is death." When production gets cheap, the work moves up, to judgment, to taste, to the next problem worth solving.

So this is the bet. The page is being rewritten in front of us, and the workspace has to be rewritten with it. You bring the judgment. Your agents bring the output. The decision becomes the record, the context becomes the moat, and the whole thing compounds into the brain of your company. If you see what we see, come build it with us.