Try It Is the Business Model

In The Feed Turns Around I argued that every personal AI app will open on a feed of activity, each item tagged IRL for what your AI did and SIM for what it would have done, with one button under every SIM card: Try it. In the section on how I could be wrong, I wrote that the whole thing depends on volume. Agents have to do enough on their own that the activity outgrows a chat thread.

I did not say how that volume arrives. Here is my prediction: it arrives through that one button. The SIM card is not a feature of the strategy. It is the strategy.

The meter and the iron

At the turn of the twentieth century, Samuel Insull ran the electric company in Chicago. He had started as Thomas Edison’s secretary, and he understood something about the business that his rivals did not. A generating plant is a huge fixed cost. It costs nearly the same at three in the morning, when nobody is using it, as it does at six in the evening, when everyone is. Every idle hour is money lying on the floor.

So Insull did not only sell electricity. He sold reasons to use it. His company sold electric irons door to door. Electric shops showed the latest appliances. Rates were cut to encourage more use, and new uses were found for the hours the plant sat quiet. Engineers call the idea load factor: how close the plant runs to full, all day long.

What matters is how this felt from the kitchen. The bill went up with every new thing plugged in. Nobody remembers it as being taken. The electric iron meant no more heating heavy flatirons on the stove. The washing machine meant no more washboard. Each step up in the bill came with a chore you could feel was gone, and people defended those bills the way they defend nothing else they pay for.

Why the SIM button carries the weight

Every personal AI company has the same problem Insull had. Its plant is a data center, and a data center is a fixed cost that hums whether you use it or not. The company needs you to hand over more. But you cannot be asked for authority in the abstract. A permissions screen is a wall of checkboxes nobody reads, and a chat that asks “should I start handling your invitations?” gets a nervous no.

The SIM card solves this the way the electric shop did. It does not ask you to imagine the appliance. It shows it working, on your own chores, with your own history: here is what I would have done this morning, and here is why. You are not being sold a capability. You are watching your own Tuesday get lighter, and being offered the rest of the Tuesdays.

And Try it is a ratchet. It is easy to tap and quiet afterward. Once a rule has run for a month without trouble, you forget it exists, not because you were tricked, but because it works. The next audition arrives. You tap again. Activity grows, the plant runs fuller, and the monthly bill climbs. That is the plan.

Here is what makes it different from every growth engine of the last twenty years.

Engagement grew by taking your time. Activity grows by giving it back.

On a social feed, growth and your interest point in opposite directions: the company wins when you spend more of your evening. On an activity feed, they point the same way. The bill rises only when something you used to do disappears. A bill like that is not resented. It is defended.

The money sitting on the floor

Think about what delegating to an AI takes today. You have to notice the chore. You have to describe it in a box. You have to decide it is safe to let go of. Then you have to check the result. Four acts of attention to save one. So almost nobody does it, and the people who do mostly delegate the work they were already thinking about, not the chores they have stopped seeing.

That is the real gap. It is not intelligence. It is friction. The value is there, waiting, in every household on earth: the bills, the forms, the reorders, the scheduling, the replies, the follow-ups. People already pay other humans for exactly this kind of relief when they can afford it, and they pay gladly, because they can feel it. The money is lying on the floor the way idle hours lay on the floor of Insull’s power plant.

And notice who writes the SIM cards. You do not. Your AI does, by noticing the pattern of your life: the email that arrives every month and gets the same reply, the text thread where you always confirm Thursday pickup, the call you return every Friday, the place you stop on the way home, the request you have typed into the box four times this year. Email, texts, calls, locations, prompts. A pattern seen often enough becomes an audition, and the audition shows up in your feed with your own history attached.

Nobody builds a SIM card. The AI builds it from the pattern it sees.

And watch how big the cards get. Early on they are small: a bill, a reorder. Once your AI has watched long enough, they become whole jobs. Be your legal advisor for the court date next month. Brief you on every new client before they walk in. Replace the delivery software you keep fighting. You never imagine those jobs, describe them or build anything. You agree to them. That is the product: an ambient partner that takes load, stress and decisions off you, one yes at a time.

Try it is the thing that picks the money up. It removes all four acts at once. The AI notices the chore, because it watched it happen. It describes it for you, on the card. It tells you its track record, so you can decide in a glance. And the checking is built in, because every rule keeps showing up in the feed as IRL. What is left for you is one tap. When the cost of handing something over falls to one tap, the only question left is whether the chore is worth more to you than its line on the bill. For a billion people, for a great many chores, it already is.

What I expect

1. Promotion rate becomes the number that matters.

   Social companies obsessed over time spent. Personal AI companies will obsess over how often a SIM card becomes a rule. What the AI auditions, when it shows it, and how it explains itself will all be tuned to earn the tap.

2. Every SIM card carries a price and a time.

   About twenty minutes a month back, about a dollar a month. The audition that shows its cost will beat the one that hides it, because you are deciding whether to plug in an appliance, and appliances come with a label.

3. The bill climbs, and people defend it.

   Monthly spending on personal AI will rise steadily for the households that delegate most, and they will be the least likely to cancel, for the same reason almost nobody cancels electricity. You do not cancel the thing that does your chores. You cancel the thing you forgot you had.

4. The auditions move to the night.

   Insull’s lesson was load factor, and data centers have the same idle hours his plants did. Expect SIM cards to favor work that can run while you sleep, priced cheaper when the plant is quiet. “Overnight” on the home screen is not just a design choice. It is the cheapest hour on the meter.

5. Unplug becomes a right.

   As standing orders pile up, customers first and regulators later will demand an itemized Hired tab with a one-tap unplug on every line, the same fight that was fought over cancelling subscriptions. The companies that make the bill legible will keep their customers.

This is the logic Charleston AI already runs on. We deliver finished work, not tools, because a tool is a promise and a finished thing is a chore gone. It is also what I do with my own AI every time I say “you decide.” That is Try it by hand: one more kind of decision becomes a standing order, the work it does for me grows, and so does what it costs to run. The question I ask about each new standing order is not what it costs. It is whether it took away something I can feel. I wrote about the felt side of that in The Notification Everyone Wants Is Done and about the habit itself in You Do It. You Decide.

How this could be wrong

Already true

The new personal agents are sold by the month, and the most expensive plans exist because some people already hand over a great deal. Every company in this race is paying for data centers that cost money whether they are busy or not. And Google has already put a Chat or Task switch under its prompt box, which is the first step from asking to delegating.

What has to happen

The auditions have to be right often enough that tapping Try it feels safe, and each new rule has to be cheap enough that nobody hesitates. If the first expensive mistake people hear about comes from a promoted rule, the ratchet slips backward, and it slips for everyone at once. And people have to let their AI see enough of their lives, the email, the texts, the calls, the places, to notice the patterns in the first place. That is a grant of trust far larger than any single Try it, and it has to be made once, early.

Where I am probably wrong

Flat pricing might win. People hate meters. They loved unlimited data plans. If the companies keep one flat monthly price, SIM still drives activity, but it builds loyalty instead of revenue, and the push to audition more turns into a push to cost less. Worse, a company paid by the action has a reason to audition busywork, tasks that run but do not matter. If that happens, the felt value breaks and Try it becomes the next dark pattern. The safeguard is the one Insull’s customers had. They could see the iron work. As for the billion people at five hundred dollars a month, if it gets that easy and spending stays at a fraction of that, then friction was never the barrier. Trust was, or the value was thinner than I think. That is the test, and I expect to pass it.

Go back to the Hired tab and its nine lines. You went looking for something to cut and could not find one. That is not a trap. That is what a good appliance feels like.

Plug in the next one.

Want a feed like this before the apps arrive? Part three shows how to ask for one: Ask for the Feed. Let It Build the Cards.