I want to tell you about the look.

You probably know the one. It's not miffed. It's not confusion exactly. It's something closer to quiet, patient concern — the kind that says "I love you dear, but what is happening to you?"

I got that look recently from both my wife and my son. At the same time. Which, if you've ever received a synchronized version of the look from multiple family members simultaneously, you know it carries extra weight.

Here's what I did.

I had this idea — genuinely good idea, I still maintain — to use AI to photograph the contents of my refrigerator and pantry, have it analyze what was in there, and roll out a week's worth of meal suggestions based on the inventory. Efficient. Practical. Solves the single most recurring negotiation in our household: “what’s for dinner?”

I showed them.

They looked at me.

My son said it plainly — "Why would we even need that."

Not a question. A statement. With a period.

I heard a line recently on The Vergecast — Nilay Patel and David Pierce, in an editorial meeting, riffing on how AI companies design products.

"I feel like we should run 'the people do not yearn for automation' as a headline."

It stuck with me. I mean, I'm using it as a headline.

Here's what they were talking about. A lot of AI products are built on an assumption that people want their lives fully automated — where friction is always the enemy, that efficiency is always the goal, that if a task can be handed off to a machine it should be. And the AI industry has largely designed around that assumption without stopping to ask whether it's actually true.

For business workflows and operational tasks — yes. Automate the repetitive stuff. Map the workflows. Build the agents. I've spent a year telling people exactly that.

But human life is not a workflow.

Human life is messy and varied and taste-driven and full of moments that only have meaning because they're slightly inefficient. The dinner conversation is not a problem to be solved. It's a negotiation, a ritual, a small daily act of figuring out what you both want. The friction is the point.

My family didn't need an AI to tell them what to cook. They needed me to be present for the conversation.

Spend enough time on social media and you'll encounter them. The people who have automated their morning routines, their meal planning, their journaling, their inbox, their sleep tracking, their weekly reviews, their goal setting, and their relationship check-ins. Who describe their life as a "system" and their days as "optimized."

I understand the appeal. I really do. I have fifty tabs open at any given moment. I know the feeling of wanting to get on top of it all.

But I've watched enough of these experiments play out to know where they tend to go. You spend so much time building and maintaining the system that the system becomes the thing you're doing. The optimization becomes the hobby. And somewhere underneath all that structure, the actual texture of your life — the unplanned conversations, the spontaneous decisions, the beautiful mess of just existing — starts to thin out.

Existence is not supposed to be frictionless. Some of the best parts of it require friction.

I'm not making an anti-AI argument. That would be a strange position for someone who uses Claude Code every day to build regional intelligence infrastructure and just published a newsletter about agent economics.

I'm making a boundaries argument.

AI belongs in the operational layer of your work. The repetitive tasks, the information processing, the workflow automation, the things that have a right answer and benefit from speed and consistency. That's where the leverage is real and the tradeoffs are worth it.

AI does not belong in the parts of your life that derive their meaning from being human and imperfect and slow. The dinner conversation. The Saturday morning with no agenda. The phone call you make just to check in. The decision you sit with for a week because it deserves that.

The refrigerator idea was a category error. Not a bad idea technically. A bad idea humanly. There's a difference.

The AI-optimized life crowd will tell you that automating the mundane frees you up to do the things you love. And they're not entirely wrong. That argument works in the right context.

But I'd push back on the premise. Most of the "mundane" things they're automating aren't actually mundane. They're the connective tissue of a life. The small negotiations and routines and conversations that, accumulated over time, are how you actually know the people you love and how they know you.

Automate that layer and you haven't freed yourself up. You've just made yourself more efficient at being alone.

The balance isn't between AI and no AI. It's between the operational and the human. Between the layer where machines genuinely help and the layer where they genuinely don't belong.

My family already knew that. They just needed me to catch up.

If it's operational — workflows, business processes, productivity, the repetitive tasks that drain your time and return nothing human in exchange — automate away. Use Claude Code. Deploy your agents. Let OpenClaw run overnight. That's exactly what these tools are for and they're genuinely good at it.

But if it has the faintest touch of personal connection — family, friends, the texts you send, the conversations you have, what's for dinner, what to watch, how to spend a Sunday — try to resist the optimization impulse. Your messy, inefficient, gloriously imperfect human-ness is not a bug to be patched. It's the whole point.

The noise out there right now is loud. Everyone's selling you a better version of yourself through a subscription. Most of it is solving problems you didn't have.

You already know how to be human. You've been doing it your whole life. Don't outsource that.

I'm still going to build the refrigerator thing. Not for my family. For clients who may actually want it.

But I'm going to keep having the dinner conversation the old-fashioned way. Inefficient. Occasionally frustrating. Completely irreplaceable. How about you? What’s for dinner?

Sat Singh builds AI apps and tools for the Coachella Valley and occasionally tries to deploy them at home. His family has opinions about this. Frequently.

If this resonates, pass it along to someone in the valley who'd benefit. This is a community project — it grows the same way communities do, one conversation at a time.