Howdy wizards,
I read something that stuck with me.
How do we make sure we keep the fun stuff for us, for the humans, and give the AI the stuff that’s less interesting and less fun?
The wording puts a sense of responsibility on the reader.
We need to make sure AI gets used for the right stuff.
Increasingly, using AI doesn’t mean one-off prompting. It means agent-run automation.
That raises two questions:
Which processes should you automate, and which do you keep manual?
Once you’ve decided to automate, which parts of those processes (if any) should remain manual?
This is my take on question 1.
I’ll answer number 2 later this week.
This is not a tutorial on how to automate (I’ll save that for later).
In the mood for something more hands-on? Today’s sponsor has a webinar tomorrow that you really shouldn’t miss:
IN PARTNERSHIP WITH OUMI
A general-purpose model doesn’t learn from your production failures, operator corrections, or the work that makes your business different.
On August 11, Oumi is launching the AI Factory that runs the complete loop: Evaluate, Synthesize, Train, Deploy, Compound. We’ll show it live on a real task and reveal three product announcements.
Join Manos Koukoumidis and the Oumi team at 10:00 AM PT / 1:00 PM ET.
❦
Deciding whether to automate or not
Before I get into explaining this chart, take a sizeable pinch of this: 🧂 (that’s salt btw)
This is how I think when I automate my stuff.
Thought it might resonate with others too.
probably my magnum opus of flowcharts
The chart asks about money first: does it actually contribute to your bottom line?
Two reasons that’s the first question:
Building an automation takes time and cognitive capacity from you. Both are limited resources for humans. The type of mental capacity needed to automate a process well is especially limited, I personally have a few hours of it per day. Even with AI handling the technical implementation, you still have to get deeply involved to create an automation that’s actually helpful.
Building and running an automation costs money. Building it costs tokens or usage credits. Running it often costs money forever (e.g. model calls, APIs, hosting, ++); everyone thought this didn’t really matter with AI, until they realised it’s getting really expensive. It only makes sense if it earns more than it costs.
Time savings are not enough. Once you start automating things, you’ll discover it’s a bottomless well. There’s an endless supply of little tasks that could save you time. Automating these indiscriminately sinks your best hours into things that don’t necessarily move the needle, and leaves you less focus for the things that do.
That’s also why I’ve made it a double step in the chart. I don’t know about you — but my ego gladly assumes any process I’m doing is contributing to my business. Else, why would I have been doing it manually up until now?
But when I start to examine the things I do in an honest way, I find the good ol’ Pareto principle harder at work than anywhere else.
Chances are, there’s a few things that move the needle in your business, and a lot of little things that just exist.
The stuff that doesn’t earn money
When you identify a process that doesn’t really contribute to your business’ growth, it’s time to get brutally honest.
Why am I even doing this?
For me, often, it’s that I enjoy it. Even though I know deep down it’s not really a money-generating activity. And that’s okay. God forbid a man has hobbies! I just make sure it’s not my main focus.
The flipside is when a task isn’t really contributing to my business and it’s boring. Let’s call these by their name: inefficiencies.
Having a meeting out of politeness, keeping a social channel alive without a single lead to show for it, the project that loses money & drains me but I’m too emotionally attached to let it go. And so on.
Those processes are ripe for killing.
If I can’t, or I don’t want to yet, I keep doing them manually. I don’t automate them.
That way, the inefficiencies in my business stay visible where they can actually be seen, and removed or transformed when their time has come. When they annoy me enough that I’ll finally toss them out of the window.
If I abstract these things away into automation I’d basically pour concrete around them. And eventually, I’d have a huge messy drawer of inefficiencies I’ve lost track of why are even there.
The stuff that earns money
This one is easy to get wrong.
Once you know something earns money the impulse is to scale that thing. And with AI, anything you can render on a screen can be scaled. That doesn’t mean you should.
For me, the deciding factor is enjoyment. If something makes me money and I enjoy it, why would I automate it?
But: I use common sense. If something I do is about to get turned into an absolute commodity by AI, and I can see it coming, I’d use that information.
Even things I like doing have aspects of them which I’m less fond of. You can automate sub-processes, so your time only goes to the highest leverage sub-process—likely the thing you’re most passionate about. I’ll go into how I identify those sub-processes in my own work in my next post.
Finally, there’s the stuff that makes money and is no fun for you. For me, it’s things like accounting, analytics, research, proofreading and more. Essential yes, and also perfect automation candidates.
❦
Right now there’s someone automating the thing you’ve decided to keep for yourself.
There’s pressure in that.
And when the numbers are bad, everything starts looking like a cost to cut. Fun or not.
Automating boring, money-generating processes is the antidote. It gives results.
And results give us freedom to do what we want, including deciding which parts of our work stay human.
This is how to make sure we keep the fun stuff for us: make the boring stuff pay for it.
What's your verdict on today's email?
See which AI use cases are paying off with Context Windows Pro
Most companies pick AI use cases by brainstorming internally. 90% of those initiatives fail.
I’ve created Context Windows just so you can pick the winners.
🟦 Find high-performing use cases from 2,000+ companies at contextwindows.ai, or book a demo with me
Disclosure: To cover the cost of my email software and the time I spend writing this newsletter, I sometimes work with sponsors and may earn a commission if you buy something through a link in here. If you choose to click, subscribe, or buy through any of them, THANK YOU – it will make it possible for me to continue to do this.