Start with the task you hate
Your first AI agent is not meant to maximize revenue, but to make you want to build the next one
I previously wrote that your first personal-agent task should be something small, repetitive, and annoying. Emphasis on the annoying.
This week, I read an opposing argument:
Annoying tasks do not bring revenue, so you should start with a task close to revenue.
That makes sense.
That said, one does not exclude the other: lead follow-up and quote preparation from discovery calls are good examples of “boring” and repetitive tasks, hence probably annoying, yet still close to revenue.
Still, I would not lead with those if I had a more painful task I would happily remove from my to-do list.
Your first agent is an exercise in learning how to delegate. The risk of starting with the dream workflow — i.e. get more leads, book more calls — is that it needs the most context, judgment before you can trust it. You will need quite a few iterations and might give up, deciding agents “don’t work for you,” when it gets hard. And it will.
The task you hate is a greater incentive to complete the setup, and it is probably better defined, particularly around where it can fail.
So for your first agent, start with the task you never want to do again.
The money-making ideas can come next, after you have learned the ropes and best practices for instructing your agent.
If you want help finding the right place to start with AI agents, without wasting time on the wrong setup, let’s talk.