A head full of ideas. Not unfinished tasks.
What helped me most was the idea that I didn’t have to keep everything in my head. Tasks, ideas and unfinished business need somewhere to go. That makes it easier to focus on what I’m doing now, without constantly remembering fifteen other things.
That foundation still works for me. What’s changing is how the work gets done. With AI agents alongside me, getting from ‘I need to do this’ to ‘done’ can involve considerably less of my own effort. Here are seven changes I see in that approach. This is my adaptation to the new tools.
Seven changes that make a difference
-
More tasks can fit into five minutes. A first draft, an offer comparison or extracting data from a file can stop taking half an afternoon. With AI document processing, an agent may have material ready for review before I finish my coffee. Naturally, that depends on the task and the quality of the sources.
-
Instead of adding another item to the list, I delegate it. If I know the outcome I need, I hand the task to an agent. In a properly configured system, the task is recorded, the agent works on it, tracks the agreed deadline and comes back with a result or a question. If the idea is still vague, I capture and clarify it first.
-
I describe the outcome, rather than every mouse click. ‘Compare these three offers, highlight the differences and cite your sources’ is more useful than ‘deal with the offers’. A few specific sentences at the start save questions and revisions later. I also agree what a finished result needs to include.
-
I organise the knowledge so I don’t have to explain the business from scratch. Current rules, decisions and materials in a single company knowledge base can give agents the context they need. They need appropriate access, and I need to know which source is current. Less copying the same information, more continuity.
-
I turn a recurring task into a process. If I delegate the same thing every week, it makes sense to define the trigger, the steps and my review point. That creates process automation that can start under agreed rules. Remembering to say ‘prepare the report’ no longer has to be my regular job.
-
I review decisions and exceptions. Instead of asking each agent how things are going, I want to see what’s ready, what’s stuck and where my decision is needed. A regular review still makes sense. It also helps check that delegated work is heading in the right direction before a small mistake grows.
-
‘Done’ needs something to back it up. I need a file, a source, a comparison or evidence of a check. When organising data, I want to know what changed and whether the data remains consistent. A cheerful ‘done!’ in a conversation doesn’t close the task on its own. I agree the agent’s scope for independent action in advance.
David, I’m looking forward to the next chapter
I still recommend David Allen’s Getting Things Done. It’s a foundation worth building on with AI agents today, for a clearer head and more tasks completed. More about the method: Getting Things Done.
Perhaps David will publish an edition that takes AI into account? I can’t wait! Especially for the chapter on a weekly review with a team that doesn’t need coffee.
Which tasks on your list could you already hand over to AI agents? Let’s talk about where to start.