All work
Case study · Napblog · Nap OS
Ninety fields became three questions. The rest waits until the work needs it.
In Ireland you need experience to get a job, and a job to get experience. Nap OS was Napblog's answer. As the design intern, I was handed the flow they had built, talked to the people stuck in that loop, and redesigned it.
01The brief
No experience, no job. No job, no experience.
01
Employers ask for Irish experience
A degree from elsewhere, or a career from elsewhere, does not get read the same way here.
02
Every route to it needs a yes first
An internship, a placement, a graduate programme. Someone has to choose you, and the reason they do not is the experience you cannot get.
03
The concept
Take the gate off. Let a person start building something an employer can look at this week, without anyone admitting them first.
Pugazheanthi Palani, the founder, handed me the problem statement, an idea of what onboarding should do, and the flow Napblog had built so far. The job was to redesign it. What the brief did not yet have was the users' side of it.
02The people
I spoke to around fifteen people at three points in the same loop.
Inside
Napblog's own interns
Already through the door, and the clearest on what had got them there and what had not.
At the door
Prospective interns
Mid-application. Every conversation came back to the same worry: what they had to show.
Outside
International graduates
Job-hunting in Dublin with a degree, a portfolio, and no way to make either count as Irish experience.
The same person at three moments. That was the thing worth noticing: the platform did not need three products, it needed one that met people wherever they had got to.
I took the initiative on this. There was no research brief; there was a founder who knew the problem and people living it, and the second group had not been asked.
What I heard
Nobody was short of things they had learned. They were short of things they could prove.
03The shape
The brief said take the gate off. The conversations added a person who reads the work.
01
A brief small enough to finish
A scoped piece of work with a named deliverable and criteria published up front. Weeks, not a semester.
02
A person who reads the work
A brief with nobody at the other end is homework. The reader is what made the work count, for everyone I spoke to.
03
A record someone else can check
Not a certificate. A record with a human behind it that an employer can ring.
The founder's concept removed the gatekeeper. The people I spoke to did not want the person gone, only the waiting to be chosen. So: take the gate off, keep the human. That was the shape I took back to the founder, and the shape I redesigned the flow around.
Everything on this page follows from it: the flow, the screens and the system behind them, designed during my internship.
04The call
Ask what the match needs. Ask the rest when the brief does.
Before
Ninety fields, opening at one percent
The profile I was handed ran to ninety fields, twenty-seven of them social links, and included a satisfaction survey before a member had done anything to be satisfied about. Almost none of it was needed to pick a first brief. Two things were: where a person is starting from, and what they want to work in.
After
Three questions, then work
Where are you starting from, what do you want to work in, which brief. Each brief says what you will have to show at the end of it. The CV, the links and the availability get asked by the brief that needs them. The profile fills itself from the work.
The founder
Fewer questions, a quicker start
He had handed over the ninety fields. His reaction to three: the first run asked less of a person and got them to a brief sooner, which was the product's own argument.
Before. My Portfolio, the first screen a new member saw: ninety fields in eight sections, one percent complete.
A profile is not the price of entry. It is what the work leaves behind.
The first run. Three questions, then a brief on the desk. It runs.
05The test
Two sessions, one task: start your first brief.
The prototype
Built to be used, not looked at
I built the first run and the workspace in Lovable, a no-code builder, so people could click through them rather than watch me. Two sessions of around three people each, Napblog's interns and people applying to join. The task was to start a first brief: sign in, answer the questions, choose a brief, then find their way around the workspace it opens into.
What held
The three questions held
People said they would not have wanted to pick much before they had seen a brief.
What did not
The charts needed explaining
The dashboards were the part people were least sure of. Most had not read a chart like this before and wanted to be told what it was showing them. So the dashboards were cut back: one question per screen, and the fewest charts that answer it. The point was to tell someone what was happening with their work, not to make them wary of the platform.
06The redesign
The rest of the flow I was handed, and what I changed.
The desktop
Before: a desktop, a dock, and every app opening as a tab inside one pane. After: windows. Two apps open side by side, the brief next to the work.
The brief
Before: ten rows of “follow the guided template”, two hours each, the criteria nowhere on the page. After: the brief pinned beside the tasks, and a task that says what you hand in and what it is read against.
The record
Before: promised, with no screen of its own. After: the one screen designed for someone who is not a member: an employer, checking.
The desktop. Two apps open at once, and windows that behave like windows.
One brief. Open a task, hand it in, watch it go to review.
The journey, in six screens. Sign in, pick a domain, read the brief, hand in, feedback comes back, share the record.
The verified record. The screen the promise is made on, seen by the employer who checks it.
07The numbers
One log, three dashboards. Each answers one question.
The member
Where did my hours go?
Estimated hours as an outline, logged hours as the fill. An overrun shows as the fill passing its outline, with the reason one hover away.
The reviewer
Am I keeping the promise?
Every hand-in is promised an answer within seven days. The reviewer's screen draws that promise as a dashed line and anything past it in amber.
The founder
Where does the cohort stall?
One dot per hand-in, the median drawn as a line. A wait that everyone shares shows up as the same empty stretch.
What we discussed
How much to show was the long discussion with the founder and the engineers: which details a member, a reviewer or the founder needs in front of them, and which would only be there because the log had them. The sessions settled it: each dashboard came down to one question.
The names and numbers on these screens are invented. The questions are the design; the numbers are not findings.
Where the hours went. The hand stops on task three, and the reason it ran over opens beside it.
The reviewer's queue. Twelve hand-ins, the seven-day promise as a line, and the two that missed it.
The cohort. Twelve members, one brief, and the empty stretch they share. The funnel and the waits are counted from the same hand-ins.
08The system
Sixteen components, so thirteen screens agree with each other.
01
One action colour
Forest for the chrome, one green for anything you can press, the bright green as fill only. Amber for anything pending. There is no red anywhere in the member flow: a revision is small and specific, so it is never dressed as an error.
02
Three faces, one job each
Inter names things: titles, numbers, people. Hanken Grotesk operates and reads: body, buttons, labels in sentence case. Plex Mono only for values someone might quote, the reference and the file names.
03
Numerals, not ticks
The three criteria are numbered 1, 2, 3 in rings, and the same numeral follows each one through the brief, the task, the hand-in, the feedback and the record. Met, the ring fills green.
Sixteen components in three families: five core (button, pill, numeral, avatar, file chip), five for forms (input, textarea, toggle, select pill, choice tile), six surfaces (menu bar, window, dock, card, ledger, modal). Tokens for colour, type, spacing and shape, and a click-through kit built from them.
Every text colour clears 4.5:1 on the surface it sits on.
The system on one sheet. Nine colours, three faces, the parts every screen is built from, and ten chart shapes with the question each one answers.
09What it became
It got built. People are using it.
My part
The research and the redesign of the flow, worked through with the founder and the two engineers who built it.
Today
Nap OS is live. Members start a brief without anyone choosing them first, hand work in to a person who reads it, and build a record an employer can check. The loop the research was built around is the one the product runs on.
10The reflection
What I would do differently.
01
Talk to employers
The record is made for employers. Next time I would talk to a few of them first, to learn what they would want to see on it.
02
Keep the words
Next time I would keep notes from the conversations, so the people I spoke to could be quoted in their own words.
03
Test for a number
I would like to run task-based tests, to measure how many people finish the first run and how long a first brief takes.
See other work
All work
This is the strongest thing you have handed in. The screenshot-then-sentence pattern makes every stall checkable in seconds. Keep that shape for the findings page.