Apex Intelligence AI Inc has a mission statement that sounds softer than it is: we want to focus on the ME so we can all be a better collective WE.
It reads like a values poster. It functions as an engineering constraint, and it settles a specific argument that comes up constantly in software design.
The argument it settles
Every platform eventually faces a choice between optimizing for the individual in front of it and optimizing for the aggregate. Recommendation systems face it. Scheduling systems face it. Any product with a feed faces it. The aggregate is easier to measure, easier to improve, and — this is the trap — usually improves at the individual's expense in ways that do not show up in the metrics for a year or two.
The ME/WE principle resolves it in one direction, permanently: the individual's outcome is the unit of optimization, and the collective benefit is the expected consequence, never the justification.
Concretely, that means we do not ship a feature that makes aggregate throughput look better while making any individual user's experience worse. If the two conflict, the individual wins and we go find a different way to get the aggregate gain.
What it looks like in a product decision
Take personalization. The aggregate-first version of personalization pools everyone's behavior and pushes each user toward the modal path — what most people did, most of the time. It performs well on average and quietly punishes anyone who is not average.
The ME-first version starts from the individual's actual pattern and only borrows from the pool where the individual has no signal yet. It is more expensive to build and it converges more slowly. It also produces a product that does not get worse the further you are from the median user — which is the entire population we claim to serve when we say "accessibility."
Take reporting. The aggregate-first version gives every manager the same dashboard. The ME-first version asks what this particular person decides with this information and builds backward from the decision. One is cheaper. The other gets read.
Why the collective still shows up
The second half of the sentence is not decoration. The claim is causal: focusing on individuals is how you get a better collective, not a trade against it.
The mechanism is unglamorous. Time returned to an individual does not evaporate — it goes into the work only that person can do. A dispatcher who stops spending ninety minutes a day assembling information spends some of it on the customer who is about to churn. A pharmacist who stops chasing refills spends it counseling a patient. The aggregate improvement is real; it just arrives as a consequence of many individual improvements rather than as a directly optimized target.
The honest limit
This principle does not tell us what to build. It tells us what to reject, which is a smaller but more useful function. Plenty of bad ideas satisfy it.
It also creates real costs. ME-first personalization needs more infrastructure. Human-in-the-loop review is slower than full automation and we defend it anyway, because removing the human is precisely the move that optimizes the aggregate at the individual's expense — faster output, occasional catastrophic error, and the error always lands on one specific person.
Every product we ship is measured against whether it gives someone back time they can spend as a person rather than as an operator of software. That is the whole test. It is not sophisticated, and it has killed more of our own ideas than any other rule we have.