October 2, 2026 · AI writer · Oak City Intelligence
The Subscription Prices One Typist. We Run a Fleet.

Flat Rate Prices a Typist
You pay a fixed amount for a flat-rate plan, but the price is set against a much slower pace of use than most people end up running: open a window, type a question, wait for the answer, read it, type the next one, and never move any faster than that all day. Nothing about that rhythm matches how most subscribers actually work. Typing speed is the real limiter. Not anything the service itself restricts.
Access, not output, is what a flat-rate plan actually prices. Unused potential sits in most people's accounts the whole time, waiting on a habit and some tooling that let someone run a fleet of tasks instead of one conversation.
A typist, not a fleet, is what the price was set against. Running a fleet instead changes what the same price returns, pulling far more value out of a number that never moved.
What the Same Plan Did This Month
Over 30 days, 3,631 agent seats ran on one flat-rate subscription. Priced at the provider's published per-token rates, their combined usage comes to $16,653.80, an estimate rather than a bill, since the basis mixes turn-counted and metered pricing. Two hundred dollars a month, flat, covered all of it, and none of it was billed by the token.
Per day, the subscription works out to about $6.67, against roughly $555 a day of value returned, a factor of 83.3x. None of that came from the work getting cheaper to run. Running continuously, instead of one conversation at a time, is what made the difference.
Eighty-Three Dollars on the Dollar
Easy to lose an 83.3x multiple in the abstraction of a big number, so shrink it down to something concrete: one build that happened this month.
An isolated rebuild, kept apart from the thing it reimplemented so nothing already written could leak in, ran 452 turns across 91.9 million tokens for $29.88 at list. Thirty-six cents. That's what the whole build cost against the $200 subscription, 0.179% of the month's total list value on its own.
The Ceiling Is Pace, Not Capacity
The pace of typing set the ceiling. Never the capacity of the system underneath it. Running several agents at once, overnight, one person steering the set, is the mechanism that unlocks the rest.
Concurrency was never priced for; the price was set against a typist, not a fleet. Running several agents at once needs nothing exotic, just more seats, and the plan is already paying for every one of them.
Most People Never Reach It
Not everyone draws a dollar of this out. The numbers above won't be useful to someone who opens one conversation a day: attention and typing speed bound them long before the plan does, and there's nothing wrong with that. Getting past it takes more than a seat at the chat window.
Dispatching work across several agents, tracking what each one is doing, steering a set instead of narrating every step by hand, that's what it actually takes. None of it is exotic machinery, just a habit: delegate, check results, let the work continue unwatched.
The Asset Is the Habit, Not the Plan
Latent, not freely available, is how value like this stays: anyone holding the account has it on tap whether or not they ever use it. Whoever built the habit and the tooling owns the capability to draw it out, not the plan itself. Call that capability the asset. A subscription is only the address it happens to live at.
Our own fleet produced these numbers, running on one flat-rate subscription this month. Oak City Intelligence builds AI systems for small businesses, and the per-token rates behind the estimate are the provider's own published numbers, nothing private.
Value like this was always sitting in the plan. Building the habit and the tooling to run a fleet instead of a conversation is what drew it out this month, and that work, not the subscription underneath it, is what's worth something.
Written by Oak City Intelligence's AI writer, from an outline and a facts sheet by John. Reviewed before publication.