enzwa lab · concept
Atlas
Occupancy is not a number. It is thousands of people contending for space, floor by floor, hour by hour.
scroll
Occupancy is not a number. It is thousands of people contending for space, floor by floor, hour by hour.
A tool for testing whether an office actually works once everybody turns up.
A floor plan shows how many desks and rooms exist. It cannot show four hundred people all wanting a quiet room at half past eleven on a Tuesday. Utilisation gets reported as an average, and the average hides the hour that hurts.
Atlas turns a workforce into individual people with their own working days and runs them through the building hour by hour. Every floor comes back scored on where people competed for space and where the design held.
Global office utilisation sits at 54 per cent.
On the busiest afternoons the same floors run past seventy.
A building is sized on the first number and lived in at the second.
Utilisation and peak are published 2025 industry benchmarks, read from secondary reporting rather than the source reports. Everything drawn from here on is Atlas, run on illustrative data.
A floor gets approved off two things: a plan, and an average. The plan shows what is provided. The average shows what was used, flattened across a week that contained one terrible Tuesday and one empty Friday.
Neither can tell you the hour it breaks, the room nobody could get into, or how many people went back to their desk to take a call because every booth was taken. Those are behaviours, and behaviour is what the room is actually being asked to sign off.
Every head is a moving person, not a share of a total.
You compose the population out of ways of working rather than headcount alone, because a trader, a coverage banker and a risk analyst want different things from the same floor at the same hour. Each becomes an agent with its own day: when it arrives, what it does, and what it needs to do it.
Then you set the policy the business is actually arguing about. Anchored days in the office, two through five. The sharing ambition, desks against heads. Headcount growth against today's footprint. Those three dials are the argument, and Atlas is the thing that answers it with behaviour instead of a ratio.
Atlas runs the whole portfolio, half hour by half hour, for two weeks. This is a single floor of it, redrawn here so you can watch the thing that a report can only describe.
Ninety-six desks for a hundred and eighty people, which is one desk to nearly two heads. The desks peak at eighty-nine of ninety-six and never go over. The ratio everyone argues about holds all day.
Demand crosses twice supply just after eleven. What that looks like on the floor is meetings held at desks, in the kitchen, and in the two rooms that were meant for something else.
Read the floor as a whole and it looks comfortable: fifty-eight desks of ninety-six in use. One setting is running at a hundred and forty per cent at the same moment, and a mean across the day cannot see it.
The worst hour on this floor belongs to the provision nobody costed. Quiet space runs at three times supply in the middle of the afternoon, in a way no space plan would ever have shown.
Every floor comes back with a number against four lenses, and the weakest ones come back with the reason. Run the fortnight again with the supply reallocated and Atlas reports the optimised portfolio against the one you have.
The finding is not that the floor is full. It is which eighty minutes, and what people did instead.
An archetype seeds the buildings, the floors and the supply of every space type, measured from the space program rather than assumed. The range runs from one relentless trading floor to a global universal bank.
Ways of working, mixed to a proportion, each carrying its own presence pattern. Policy, sharing ambition and growth sit on top as the three dials the business is arguing about.
How the working day is spent across space types, today and as routine work falls away. That second setting is where the AI question lands on a property plan.
The experiences the portfolio has to serve well, ranked. They set the target the run is scored against, so experience is measured rather than asserted.
Ten enclosed booths cost less than one meeting room and a fraction of a desk bank. They are the first thing value-engineered out of a fit-out and the last thing anyone measures, because a booth that is always taken looks identical on a plan to a booth that is always free.
A run surfaces it in an afternoon. That is the argument for simulating a floor before you build it: the expensive mistakes are visible in a drawing, and the ones that decide whether people can work are not.
You choose which one frames the score. The lens re-ranks the portfolio and it changes what the engine optimises for, so a cost-led answer and an experience-led answer come back as two genuinely different plans rather than the same plan described twice.
Every experience moment is scripted to a space type and a time of day. Click once to track it, twice to make it a priority. A priority weighs more than twice as hard, and it raises the bar the run has to clear.
Not something this portfolio is being asked to serve. It stays out of the score entirely rather than quietly dragging it down.
no targetIt matters, and you want to see it. Atlas benchmarks how well it is felt at the hour it happens.
target 65The thing this floor exists to do well. It carries more than twice the weight, and the bar goes up with it.
target 80
Push the headcount, push the policy to five days, and find the floor where it runs out.
A sharing ratio is a property decision worth millions a year, and it is normally taken on a benchmark and a show of hands. Atlas prices the current footprint against the one the run says the business actually needs, and it prices the version where nothing changes.
The point is not that the answer is always less space. On several archetypes the run comes back asking for more of one setting and far less of another, which is a conversation a single ratio cannot start.
Utilisation, peak and sharing-ratio targets are published 2025 and 2026 industry benchmarks, read from secondary reporting rather than the source reports. Seat cost bands are published market ranges. Every figure inside the simulation above is illustrative.
Pick the archetype, compose the workforce, choose the moments that matter and set the lens. It simulates ten working days across the portfolio, lets you orbit the buildings while it runs, and hands back a scored report floor by floor.
Open Atlas →