Most of us know the Eisenhower Matrix: the 2x2 grid for categorising urgent and important activities. It’s been a mainstay of productivity thinking for decades.
But, the nature of work is shifting; more people are working for themselves, embracing portfolio careers or taking on fractional roles across multiple organisations. In the UK and Ireland fractional working is gaining serious momentum. At the same time, broader research shows a rise in flexible, self-directed working models and a growing preference among workers to carve out more autonomy in how, where and when they work. For solo operators and independent professionals this means our management tools need to catch up; we’re no longer managing an employee workload, we’re managing our own entire capacity.
As a fractional, solo operator, myself, I don’t have a team to delegate to. Which means that anything that would naturally fall into the “delegate” box ends up being done by me anyway. I end up with a growing pile of “urgent but not important” tasks that hang around and never get done, as more and more urgent and important items arrive, until they eventually become “important” and have to be dealt with in a panic.
There’s also the mental burden of having these un-delegated tasks. Despite having a system, reminders and reflective times in the calendar to re-visit the backlog, I find it almost impossible to ‘let them go’; to ‘trust the system’. But that could be just me.
Re-thinking Delegation
So I started to think about what it means to have items for delegation, but no-one to delegate them to. The box on in the matrix is not doing any heavy lifting; it’s just a dump for things that have to be done (because they’re urgent), but also have to wait (because the urgent and important things need doing).
Most of those un-delegated tasks are not hard; they are either repetitive, fiddly, or too small to feel worth starting. So I began handing them to AI tools - custom GPTs / RAGs / Agents. I just ‘chuck them over the fence’ to AI and get on with the more important stuff, and then review the output later (as I would have done, had I delegated them to a person - at least until I had confidence in them).
It’s not going to be possible to delegate every such activity to AI; and certainly not necessarily ‘as is’. There is a bit of thinking required to make sure that what you want can be done this way, which part of parts of it are suited to such delegation, and how you’re going to pull it all together at the end.
So I’ve renamed the ‘delegate’ quadrant as ‘deliberate’. I determine if the activity can be done by AI (as a whole or in part), hand it off and piece it all back together again at a later time.
For solo and fractional workers, the ‘deliberate’ box becomes the closest thing to extra capacity; over time you build out a collection of AI personas that can stand in for your non-existent team.
By way of an example, for my creative writing hobby/sideline, I have three custom GTPs - a project manager, a marketer/publishing expert, and a ‘sounding board’. These assist me in working on outlines for posts, and critiquing my writing, amongst other things.
The benefits I’ve noticed:
- Less mental clutter – far fewer “shoulds” taking up space.
- More focus – & more time for strategic and creative work.
- Consistent progress – even small tasks move forward without manual effort.
- Increased headroom – a sense of capacity without expanding the team.
Where it doesn’t work
AI delegation doesn’t suit everything. Tasks that depend on empathy, negotiation, or nuanced judgement still need human insight. And AI outputs always need reviewing. I think of it as an assistant rather than a replacement for a real person.
From delegation to automation
If you find yourself handing off similar jobs again and again, create an automated workflow: pre-set prompts, templates, or integrations that let the task happen with minimal friction. It turns delegation into a background process.
Another example from my creative writing. Once I’ve published a new post on The Ode Map, an agent automatically creates a stack of social media posts about it. Once I’ve vetted them and amended, approved or rejected them, they enter a ‘pot’ of posts that get scheduled by another agent. I use AirTable to hold the copy and n8n to run the agents and automations.
By minimising the time I need to spend on doing these activities, I can spend more time on the deep work - the thinking, creating, problem solving; the things I want to spend my time on (and that I wouldn’t delegate to anybody, let alone AI).
Conclusion
We’re all learning what it means to lead and deliver in a world where the “team” increasingly means fewer people and more technology. The next evolution of productivity frameworks won’t be about replacing human effort, but about understanding where it matters the most. The more consciously we integrate AI and automation into our activities, the more we can effectively shape & orchestrate work, both for ourselves and for the systems we build to help us.