Over time, the system gained more statuses, versions and places to look for information. I started spending time on coordination again.

A review revealed, among other things, that deadlines were not visible where I needed them, and that texts waiting to be rewritten were hard to find again. The information was stored. Yet the next action was unclear.

So I kept working towards a shared, up-to-date data foundation and a simpler way into whatever needed attention. “My Day” was meant to show what I had to read, decide on or follow up.

That also changed how I evaluated new features. A feature had to help me get through a task. If it demanded more data entry, searching or checking, it had to be reconsidered.

This is where systems thinking and UX met in a practical question: How much of the context does the user still have to hold in their head?

Approval needs solid ground

From the start, I kept the decisions with me. AI could research and suggest. I would assess, approve and send.

As the work went on, it became clearer that the approval itself also needed support. A well-written suggestion is hard to assess if its basis is unclear, or if earlier corrections have been lost along the way.

So I worked on how updated clarifications, documented experience and feedback should carry through the work, across assistants and work sessions. It had to be possible to tell what was still uncertain from what had been settled.

That gave me a new question to ask of the workflow: Do I have enough insight to make a qualified judgement here?

In Moral Crumple Zones, researcher Madeleine Clare Elish describes how people can be held responsible for the actions of automated systems over which they had limited control. Her research concerns different and more serious situations than my platform, but it points to a relevant difference between responsibility and actual influence.

For me, this means control has to be designed into the whole process. The user needs the chance to understand, correct and influence before a decision is made.

Expertise sets the direction

My experience in brand, strategy and design shaped the work long before I evaluated the finished result.

The strategic perspective helped me choose what the solution should prioritise. The UX perspective drew my attention to what the user had to understand and remember. My understanding of brand turned voice, precision and visual quality into practical requirements.

One example was documents that strayed from approved visual references. A general description of colours and style left too much room for interpretation. I had to anchor the work more firmly in the actual references and check the finished deliverable.

I recognise this from brand implementation. A direction gains value when it can be followed through the many small choices that shape an experience. That takes usable references, clear expectations and a workflow that supports quality.

It is also a reason to involve subject-matter experts early when adopting AI. They know the judgements, exceptions and quality standards the solution has to handle. That knowledge should shape what gets built and how the work is divided.

Building became a way of learning

I moved back and forth between exploring the need, scoping the task, developing and testing. The Double Diamond gives useful language for this movement, but my experience did not follow a tidy order. Using the platform could send me back to the problem definition itself.

Being able to build made it easier to challenge my own assumptions. I could try an idea while the need was still concrete, see where it fell short and change it.

Hyper Island’s Learning Spiral gives useful language for how experience turned into learning: doing, reflecting, drawing out insight and applying it in the next attempt. Each round builds on the last.

Illustration adapted from Hyper Island’s Learning Spiral.

When the deadlines were stored but hard to find, I had to rethink what an overview should mean in practice. This is also a way of working I know from Emerging Tech at Hyper Island. Curiosity and play make room to experiment. Experience provides the basis for prioritising.

AI shortened the path to something I could examine. Deciding what deserved to be taken further still required professional judgement.

Double Diamond diagram from Jason Carlin’s free Figma template via UIKit Free, based on the Design Council’s Double Diamond.

The gain has to be weighed across the whole workflow

A faster draft can save time. The gain shrinks if checking it means redoing all the research.

So I also began to weigh the cost of automation: follow-up, corrections, maintenance and resource use. Among other things, I reduced how often the system checked whether new contacts were needed, because more frequent checks did not give me a matching benefit.

As early as 1983, in Ironies of Automation, Lisanne Bainbridge described how automation could create new challenges for the people left to handle what remained. That perspective helps me ask a concrete question: What work is the user left with?

My background in sustainable change makes this part of how I evaluate the solution. I care about resource use, the load on people and what it takes to keep a new way of working useful over time.

For this platform, the total time saved is still to be measured. I want to examine the gain across the whole process, including the work that comes with the technology.

When others use the solution

I use the platform myself. Interest from people around me has led me to start user testing and set up a tailored version for others.

Now the questions are changing. I know the logic because I helped develop it. Others need to understand the workflow without knowing its history.

So I am exploring what is self-explanatory, where support is needed and which needs are genuinely shared. The feedback will also help me see what should be simplified before I add more.

This is where the change perspective becomes concrete. A working solution needs to find a place in people’s habits, make them feel confident using it and be worth the effort it takes to learn.

What I take with me

The career platform started with a personal need and has become a practical development project where I use strategy, design and technology to solve a concrete problem.

The most important lesson is how closely problem understanding, priorities and execution are connected. The quality of a solution is shaped by the choices we make before we build, by what we discover through use, and by the willingness to change direction when experience calls for it.

I see clear value in this for brand and marketing work. There too, ambitions have to become priorities, insight has to carry through the process, and people need to be able to deliver a quality they understand and can stand behind.

This is the work I thrive in: understanding the whole, giving a direction its shape and staying close enough to the execution to learn from what actually happens.

References

Madeleine Clare Elish: Moral Crumple Zones, 2019.

Lisanne Bainbridge: Ironies of Automation, 1983.

Hyper Island: The Learning Spiral and experiential learning.