Having a long day?Pink noise, even across every octave with a slow 10 Hz swell, takes the edge off the room. Attention lasts longer when less is competing for it.

Human Factors · Human-AI Interaction · Ethical Design · Concept

Built to calm the fear. Realised the data was the threat, not the answer.

Role
Behavioral Researcher & Designer
Scope
Deep research · model prototype
Type
R&D Human Factors · HAI
Year
2025
Status
Concept

38%

feared automation would replace their job within two years

ManpowerGroup Global Talent Barometer, June 2025 ↗

2026 update43% — ManpowerGroup, January 2026 ↗

32%

expected AI to leave them with fewer job opportunities in the long run

Pew Research Center, February 2025 ↗

2026 update49% personally afraid of losing their job to AI — Modern Health, April 2026 ↗

22%

strongly agreed their job was safe from elimination

ADP Research, 2025 Global Workforce Survey ↗

FieldworkJuly–August 2025, 39,000 workers across 36 markets — published March 2026 ↗

Being afraid for your job changes how you read. Attention narrows, working memory fills, and the sense that you still get to decide starts to thin. That is the state these three numbers meet you in — and most career tools answer it with more of them.

Mindure is a career guidance dashboard concept for people worried that AI will take their job.

I researched the problem and redesigned an AI-generated draft of its six sections for people under stress.

View prototype

Career guidance matched to how your mind is wired.

Mindure matches how a person’s mind is naturally wired with the live job market and the skills each role needs, so they can choose work that fits and is harder for AI to replace.

Six sources feed one view: nervous system profiling, real-time economic data, job market updates, upskilling, strategic planning, and wellbeing.

Each section shows only what a person can act on now, and says in plain language where its data comes from.

The design problem was timing: career data arrives exactly when a person is least able to take it in.

Four fields studied; stress set the rules.

Across the four fields, three findings about stress, from human factors research, became constraints the interface had to meet.

Resource links

Three findings about stress became rules.

Each one rules something out. If attention narrows under threat, urgency cannot be used to prompt action. If load has a ceiling, completeness stops being a virtue at the point the ceiling is reached. If automated judgement erodes autonomy, a score is a verdict dressed up as an insight.

Together, they decided more of this interface than any of my preferences did.

What does stress do to how people take in information?

Anxiety narrows attention

Under threat, the brain processes less.

Cognitive load has a hard ceiling

Go past it and the brain shuts down.

Automation anxiety erodes autonomy

It erodes the ability to make your own decisions.

Career signals overflow one person’s capacity.

Career anxiety doesn’t come from a lack of motivation. It comes when an unstable market, a person’s limited capacity and high-stakes decisions all meet, and the person can no longer make sense of it.

When signals don’t connect, users face stress and uncertainty.

In the drawing below, each dot is a signal arriving. The box is what one person can hold and act on at once, and it is already full. That much volume outside it is normal. What fails is the assumption that someone will sort it.

Why are even capable people struggling right now?

Career anxiety as the overlap of three pressures Three circles of equal size overlap. External volatility and internal capacity sit side by side at the top; decision pressure sits below and between them. Career anxiety is the small area where all three meet — it is what the three produce together, not any one of them on its own. External volatility Internal capacity Decision pressure Career anxiety

Why don’t career signals add up?

Market signals

Economic signals

Personal signals

Planning signals

How much can one person hold at once?

  • Planning pressure
  • Personal capacity
  • Economy & automation
  • Market & skills

Dots inside → manageable loadDots outside → overload

Sense-making under uncertainty A field of dots in four series — planning pressure, personal capacity, economy and automation, market and skills. A rounded box in the middle is packed full of them: that is one person’s workable load, already at its limit. The dots scattered beyond the box are the signals that arrive anyway and have nowhere to go — the overload.

It shows only what you can act on now.

Macroeconomic signals, skills-gap mapping and stress-state detection feed a single output layer.

Six sources, each with limits.

Every section also tells users where its data comes from, in plain language, at the moment they need it. It sits on the screen itself.

Each section has a job and a line it must not cross: Market and economy gives context but never urgency, and Career actions shows options but never tells the user what to do.

How might six sources speak with one voice?

The multi-modal model: six sources, one intelligence Six sources sit around a single centre and each connects to it: nervous system profiling, real-time economic data, job market updates, upskilling and learning, strategic future planning, and wellbeing and resilience. They meet at multi-modal system intelligence in the middle — the point being that no source is read on its own, and none of them reaches the person directly. Job Market updates Real-time economic data Nervous System profiling Wellbeing and Resilience Strategic Future planning Upskilling and learning Multi-modal System Intelligence

Where does each section have to stop?

SectionResponsible forNot responsible for
OverviewImmediate orientationPredicting outcomes
Personal InsightsDescribing thinking patternsScoring or ranking the user
Career ActionsSurfacing options with reasoningTelling the user what to do
Market & EconomyContext for decisionsTriggering urgency or fear
Skills BuilderLearning tied to your profileImposing a fixed sequence
Strategic PlanningQuarterly directionDaily commands or schedules
Dashboard responsibility map showing each section’s permitted scope and constraints
The original responsibility map. Open it full size.

Redesigned the AI draft for anxious minds.

Each redesign traces back to a human state.

One place to start.

Ten scored modules became five cards, with one thing to look at first.

Overview: where do I start?

Read the observations and decisions below

AI-generated dashboard overview showing ten scored modules simultaneously Raw AI output Human-in-the-loop

Raw AI output

  • 10 scored modules displayed simultaneously. No visual hierarchy, no entry point.
  • With no single place to start, the brain can’t begin, so it freezes.

Decision paralysisCognitive load

Human-centered redesign

  • Five cards, with one thing to look at first.
  • Because there is one way in, there is less decision paralysis. Volatile data stays hidden until the user is ready for it.

Progressive disclosureSingle entry point

Strengths in plain words.

Percentage scores on inner traits became plain-language strengths, compared only with the person’s own history.

Personal insights: who decides who I am?

Read the observations and decisions below

AI-generated personal insights showing percentage scores on cognitive traits Raw AI output Human-in-the-loop

Raw AI output

  • Percentage scores on cognitive traits: 94% spatial reasoning, 87% pattern recognition.
  • When a system scores someone’s inner traits, they start to doubt their own sense of who they are.

Social comparisonAutomated scoring

Human-centered redesign

  • “How your mind works”: strengths in plain language, compared only with your own history.
  • Because you are compared only with your own past, there is nobody to compete with.

Self-referential framingPlain language

Options with their reasons.

A schedule of commands became live job options that show their reasoning, so the decision stays with the user.

Career actions: who decides what I do?

Read the observations and decisions below

AI-generated career actions showing prescribed tasks with exact times and effort labels Raw AI output Human-in-the-loop

Raw AI output

  • Actionable roadmap with exact task times, effort labels, deadlines, and a cognitive load tip telling you when to work.
  • Commands take away the sense of control, and that sense of control is what most shapes how anxious a person feels.

Prescriptive automationLoss of perceived control

Human-centered redesign

  • Live job postings that show their reasoning (“why it fits your thinking style”), with an ethical boundary that stays in the sidebar.
  • The system shows options and its reasons, so the decision stays with the user.

Perceived controlVisible reasoningEthical boundary

Market data as calm context.

A live ticker stamped “2 min ago” became stable background that the user opens when they choose.

Market and economy: is this urgent?

Read the observations and decisions below

AI-generated market dashboard showing live economic indicators updating in real time Raw AI output Human-in-the-loop

Raw AI output

  • Live economic indicators (GDP, unemployment, inflation, the tech sector), all updating in real time and stamped “2 min ago.”
  • A live ticker reads as a threat, and a threat pulls attention away from planning and toward scanning for danger.

Threat attentionManufactured urgency

Human-centered redesign

  • Market Intelligence: the same data, framed as calm background for decisions.
  • Because the data is framed as stable context, the user decides when to look at it, and nothing pulls them in.

Context framingUser-initiated engagement

A learning path that fits.

A fixed sequence became a path tied to the person’s profile and direction, with its reasoning visible.

Skills builder: who sets the order?

Read the observations and decisions below

AI-generated skills builder showing a rigid linear learning sequence with no visible reasoning Raw AI output Human-in-the-loop

Raw AI output

  • Linear learning path with a fixed sequence and no reasoning visible.
  • A fixed order undermines both competence and autonomy, and people need both to stay motivated while learning.

Imposed sequenceAutonomy loss

Human-centered redesign

  • Smart Skills Builder: learning tied to your cognitive profile and career direction, with the reasoning visible and progress measured against your own goals.
  • Because the path follows who you are, the order is yours, and progress is measured against your own goals.

Competence and autonomySelf-referential progress

Plans by the quarter.

An hour-by-hour calendar became quarterly blocks, with stress-regulation practice next to the courses.

Strategic planning: how far ahead helps?

Read the observations and decisions below

AI-generated planning showing a daily calendar with exact time blocks commanded by the algorithm Raw AI output Human-in-the-loop

Raw AI output

  • A daily calendar with exact time blocks, where the algorithm sets the schedule hour by hour.
  • Rigid plans increase anxiety when disrupted. Certainty that breaks is worse than no certainty at all.

Over-specificationBrittle certainty

Human-centered redesign

  • 2026 Activity Roadmap: quarterly blocks, loosely framed, with stress-regulation practice next to the courses.
  • Planning by the quarter gives a direction without fixing every step, so the person gets long-term clarity without short-term pressure.

Coarse time granularityStress regulation

Less on screen, more a person can act on.

For someone already overwhelmed, showing everything works against clarity. The fix is a different idea of what the system is for.

Next time: start with stress research.

I designed the calm state first and worked backwards. Next time the anxious state becomes the main design surface.

“A system that gives you everything it knows is not a helpful system. Helpfulness is knowing what to withhold, and when.”