When you want the chili, you make the chili.

Even if you’re jet lagged after a week out of town. Even if you have too much to do on a Sunday and too little time.

The stomach wants what the stomach wants.

And the brain knows how to use technology to get it.

So I cracked open a Claude conversation, gave it my never-miss recipe for context, and shared my plan to convert it into a veggie-only dish.

How many portobello mushroom caps to replace a pound of 80/20? Could I add kale? How to cook the caps?

With one chance to get it right, with four dinners for the week hanging in the balance, how could I ensure success?

By knowing where to draw Ann Handley’s line. By knowing how much AI to use to add efficiency and how much the evening deserved more of me.

Taking The Long Road

In her lovely, fortnightly email newsletter, Total Annarchy, Ann wrote last week about a four-year-old on his first flight and a magical note card connection. Later in the post she described how, to her:

“…the ultimate life hack is to know exactly when to take the long road. It’s a business edge… but also the ultimate human edge…. The magic is in the selectivity. It’s about looking at your day and saying, ‘What can I do more efficiently? And also: what deserves more of me?’”

As AI labs spit out personal agents and more powerful models rushing us towards an ever more automated future, deciding where to take our time is becoming more important each day.

And there’s a tension there, right?

We know that maybe we could do a thing faster. But, instead, deciding we need more of ourselves present more often almost becomes an act of defiance.

So, on Sunday I struck a blow for humanity.

I made the chili. Slowly.

Always Check The Answers

I made it slowly only after verifying every response Claude provided. Gut checked the recommendation to grill the mushrooms – deeper, smoky flavor is always the right call – and discarded the suggestion to add baby spinach.

Spinach? In chili? Ew.

I do the same with anything AI provides, as should you.

Another example is prep I’m doing for a training this week. Claude generated a run-sheet for the session which I painstakingly checked against what the platform settings contained.

Good thing I did because one of its assertions about two-factor authentication proved to be wrong. Had I not checked everything, it would have bitten me during the call.

When I speak with a client, it’s my name and reputation, not Claude’s, on the line as Mark Schaefer pointed out in a RISE Community conversation recently. For me, those stakes are too high to trust to a probabilistic machine with no skin in the game.

No skin at all, for that matter.

Own The Output

Anything you receive from an AI needs to be treated with a healthy skepticism. Click through to the linked sources. Do your own research.

AI can seem like an all knowing, inexhaustible well of answers. Its confident, even delivery beguiles us. It must know what it’s talking about, right?

Wrong.

The only ones who can decide are you and me. Us. We get to decide where to draw Ann Handley’s line.

The chili turned out great balanced with a dash of AI and a heaping cup of me.