An ode to high quality admin, in the age of AI.
Invisible work is the quiet backbone of every team. Invoicing that arrives on time. The weekly newsletter someone edits without praise. The formatting of a proposal that lands a new donor. These tasks feel invisible, but if they are neglected or done poorly, organisations grind to a halt.
And now that work is more undervalued than ever, thanks to the rise of AI tools. We have high hopes for AI agents and even existing tools that can write, analyse, schedule – taking these tasks off our busy minds. This sounds good in theory, and I am in favour of freeing our brains to focus on brilliant and creative things.
But many of these tasks are learning opportunities for early-career professionals. They help people build judgement, understand organisational rhythm, and grow into strategic roles. Flawless administration is a building block of any career: attention to detail, accuracy, and even yes the grit to keep plugging away at boring tasks – these are all assets I still rely on 25 years into my working life.
Let the computers do it?
This year’s job market shows this shift with hard numbers. AI has already reduced opportunities in types of work once taken for granted. A recent global estimate suggests 85 million jobs could be displaced by AI and automation by the end of 2025. Administrative support and data entry roles have seen significant effects, with hiring rates for these roles down roughly 45% since 2022. According to McKinsey research, roles most exposed to AI saw a 38% drop in job postings (higher than the average of 31%).
These global labour shifts are happening now because routine work has become easier to automate.
Fewer entry-level and graduate jobs
These changes matter most at the start of a career. In the UK, listings for graduate roles fell roughly 33% in a single year, dipping to their lowest point in almost a decade. And graduate hiring data tell a similar story: for some sectors (like tech), entry-level opportunities continue to contract sharply. One industry group found that tech graduate roles dropped 46% in 2024, with even more decline expected by 2026 as tasks that once formed the training ground for new hires are automated.
This contraction affects the type of work that young professionals do first. Where once graduates learned on the job by doing repetitive but essential tasks, they now face fewer chances to grow into roles that require nuance, judgment, and strategic thinking.
Why repetitive work is strategic work
Here’s the paradox: routine tasks prepare people for complex ones. Writing the weekly newsletter helps a person learn your voice. Managing a leader’s diary teaches empathy as well as attention to detail. Formatting a document is an opportunity to curate a piece of work that reflects excellence in its every corner.
When AI does that routine work, it doesn’t mean that humans magically get better at judgement or strategy. It removes the soil in which those skills grow.
That’s especially risky for nonprofits. Junior team members learn by doing (especially when training and coaching budgets are lean). They learn by getting tracked changes all over their draft, taking a deep breath, and making the next draft stronger. (Just like we all do. There is no position of seniority in the world where there are no more tracked changes on your drafts.)
We should automate tasks, and we should lean on AI to help us make work-life more efficient and fun. But if we are going to replace junior-level tasks with automations, we then need to redesign roles to give people the experience they need to deliver at a high level. If we don’t do that, today’s career starters will end up doing less human work and learning less. Not the direction we want to head.
What nonprofits can do
There isn’t a simple fix, but there are intentional steps:
- Treat routine tasks as training ground
Protect time for people to learn fundamentals before outsourcing them to tools.
- Pair AI with mentorship
Let AI handle repetition only when a human can still use the output to learn deeper skills.
- Track outcomes, not outputs
Focus on what a team member learned as well as what work got done.
- Expose juniors to client and strategy work early
Invite them into conversations where interpretation and human judgement matter.
Invisible work is still visible in results. I genuinely believe: when newsletters go out on time, when donors get thanked, when invoices are corrected before issue, that’s not busywork. That’s the foundation on which strategic impact is built.
And if AI changes how we do that work, we should still make sure people can learn from it, not be replaced by it. If we don’t, we risk shrinking the pipeline that nonprofits depend on to lead the next generation of mission-driven change-makers.
If these questions are already surfacing in your organisation, our services page sets out how we work and we’d welcome a conversation about what this could look like in practice.
