"Seat" pricing assumes a specific unit: one human, logging into one account, deciding once a month whether their seat still earns its fee. Agents don't map onto that unit at all. An agent fleet might be one long-running process or a thousand short-lived instances spun up by a loop and torn down a minute later - none of which correspond to a person occupying a seat. Pricing by seat has nothing to count.
There's no subscription and no seat anywhere in the core per-call product: drafting, sending, receiving, and claiming a name are each priced and paid independently of who or what is calling, with no account required to call any of them. Pay-per-call pricing is the mechanism; "no seats" is what that mechanism implies once you follow it all the way through - if price attaches to the action instead of the account, there's no account-shaped unit left to charge for.
This goes deeper than billing. There's no login, no user table, no row an identity would even need to exist in for a "seat" concept to attach to. An agent's inbox is ownerless by design - DimeBox doesn't record which wallet paid to create it. You can't bill per seat when there's deliberately no record of who the seat would belong to.
DimeBox's wallet supports an optional monthly subscription, but it doesn't gate access to anything or correspond to a person - it funds a wallet and waives a one-time setup fee for callers who know they'll be topping up regularly. See how that works. Every endpoint is callable with or without it; subscribing doesn't unlock a feature, it just changes the economics of funding the wallet that pays for calls you'd be making either way.
Most email-sending tools price per seat, per month, with a volume allowance bundled in - a model built for a human marketer or support rep deciding whether the tool earns its monthly fee, then sending however many emails that plan allows. That model doesn't have anywhere to put an agent: the agent isn't the one deciding whether the subscription's worth it, and the volume allowance is sized for human sending patterns, not a loop that might send once a year or once a minute. See a side-by-side comparison of subscription email tools and pay-per-call pricing for where that mismatch actually bites.
This isn't a pure upgrade - seat-based and account-based pricing enable things that pure pay-per-call doesn't: one invoice covering a whole team's usage, a single login where someone can see everything a group of agents has spent, role-based access to who can create or fund what. DimeBox's closest equivalent is linking a wallet to an account without surrendering the agent's own access to it - useful for visibility and management, but it's an addition on top of a no-account core product, not a seat system retrofitted onto it.
In practice, most teams running more than one agent still want some shared visibility into total spend without reintroducing a login wall for the agents themselves. That's a reporting problem, not a pricing problem - see the full lifecycle of a wallet, from checkout to ledger for where that kind of visibility actually lives today.
Consider what a seat-based system would have to do to handle a batch job that spins up two hundred short-lived agent instances to process a queue of tasks, each one making a handful of calls before exiting. Is that two hundred seats for the hour they ran? One seat shared awkwardly across all of them? A seat-based biller has no good answer, because the unit it's counting - a logged-in human session - never existed in the first place. Pay-per-call sidesteps the question entirely: each of those two hundred instances is just a sequence of priced calls, counted and billed the same way whether it ran for one second or one year.
The same logic applies to claiming a name. A vanity inbox name is purchased once, by whichever wallet happens to be paying at that moment, and then belongs to the inbox - not to an account, not to a seat holder who might later leave the team. There's nothing to transfer when "the person who set this up" moves on, because no person was ever the thing being billed.
The tradeoff is consistent with everything else about pricing this way: you get pricing that fits a machine caller making its own small decisions, at the cost of the aggregation and reporting conveniences that seat-based billing was actually built for. For a fleet of agents transacting on their own, that's the right side of the tradeoff to be on.