Designing a calm back office (and closing the loop with Fleet)

Most back-office software feels like a filing cabinet that learned to send notifications. You open it to do one small thing, approve an absence, find a contract, lock a pay run, and it answers with twelve tabs, four badges and a red dot you cannot make go away. We are Elevera Studio, a small software team in Zagreb, and we spent a long time inside tools like that before we decided to build the opposite. This is the story of how we tried to make a back office that feels calm, and why we wired it directly into Fleet by Elevera.

People-ops sprawl is a design failure, not a feature

The standard HR stack grows by accretion. You start with a spreadsheet for employees, add a separate tool for leave, bolt on a payroll exporter, buy a compliance tracker after an audit goes badly, and stitch them together with exports and a shared inbox. Each tool is reasonable on its own. Together they are a swamp. The same employee exists in five systems with five slightly different spellings, and nobody is sure which one is true.

That sprawl is usually sold as power. More modules, more dashboards, more configurable everything. In practice it pushes the hardest work onto the person least able to do it: the office manager reconciling numbers at month-end, the founder approving leave on a phone, the new hire who just wants a payslip. Sprawl is not richness. It is a design failure that got shipped.

So the first decision behind Office was a refusal. One source of truth for the whole organisation: employees, departments, employments, documents, hours and pay all reading from the same record. When a name changes once, it changes everywhere. Eight modules share that record instead of each keeping a private copy.

Calm is a feature you have to choose on purpose

Calm does not happen by leaving things out at random. It happens by deciding, screen by screen, what earns a place. We treat the interface like an editorial page: generous whitespace, one clear thing to do, plain language instead of jargon, and no decoration that does not carry meaning. A pay run should read like a document you can trust, not a cockpit you have to learn.

That editorial instinct shows up in small, deliberate choices:

• A pay run moves through states you can see: built from real hours and approved absence, reviewed, locked, exported. No magic, no surprises after lock.
• Compliance surfaces what is missing, an expired licence, a lapsed certification, before it becomes a problem, instead of burying it in a report nobody opens.
• Onboarding is a role-based checklist that actually finishes and provisions access when it does, rather than a wiki page people forget.
• Self-service gives people their own profile, payslips, documents and requests on web and mobile, so the office manager is not a human API.

We had a rule on the wall while building: if a screen needs a tooltip to be understood, the screen is wrong, not the user.

Calm also means least access by default. Every organisation is isolated at the row level, roles are least-privilege, and the audit trail is append-only and immutable. An external accountant gets a token-scoped, read-only view, enough to do the books, nothing more. Quiet software and careful software turn out to be the same software. You can read how we hold that line on the security page.

The closed loop with Fleet

Here is the part we are proudest of, and the reason Office exists at all. We already make Fleet by Elevera, a platform for running vehicle fleets. The two products share one spine, so the people side and the operations side stop being separate worlds. We call it the closed loop, and it runs in both directions.

Hours flow in. When a driver works a shift in Fleet, those hours arrive in Office as time entries, ready to feed a pay run. No re-keying, no exported timesheet emailed at month-end, no argument about whose number is right.

Governance flows out. Compliance in Office governs who can be scheduled in Fleet. If a driving licence or a permit has lapsed, that person is not road-ready, and the schedule knows it before a dispatcher does. The rule lives in one place and is enforced everywhere.

And the whole lifecycle joins up. Recruiting turns a candidate into a hire, onboarding provisions their access, and on the operations side they become a road-ready driver automatically. When someone leaves, offboarding stands that access down just as cleanly. Recruit to offboard is one continuous thread, not a handoff between systems that quietly drops people. We go deeper on the mechanics on the integration page.

The trade-offs we made on purpose

A calm product is mostly a list of things you decided not to build. That is uncomfortable, because every omission is someone's favourite feature. We made our peace with it. Office prepares payroll cleanly and exports to DATEV, a Croatian format, or generic CSV, but it is not a full payroll bureau and does not pretend to file on your behalf. It tracks performance with signals drawn from the actual work, but it is not a goals-and-OKR cathedral. It does a focused set of things and does them so they hold up under an audit.

We say no to feature sprawl the way a good editor says no to a paragraph: not because it is bad, but because it would dilute everything around it. The cost of every feature is not just building it, it is the attention it takes from every screen it touches forever. We would rather ship eight modules that feel like one product than twenty that feel like a merger.

That is the whole thesis. People operations is high-stakes, lightly resourced work, and it deserves software that respects both the stakes and the resource. Calm, editorial, one source of truth, least access, and a real loop with the operations your team actually runs. If that sounds like the back office you wish you had, read more about who we are, see what it costs, or just come talk to us.