The commercial model

Two ways to pay. One product underneath.

A small business and a developer want opposite things from a bill. The business wants to know the number. The developer wants to pay for exactly what runs. The mistake the industry made was forcing both onto the same model, and then forcing a product change to move between them.

What is not in the bundle. Mail is not offered at launch. Domain registration is not resold; bring your own and the pod points DNS at it. Both are candidates for a paid add on later, and neither is load bearing for the product to make sense.

Nothing on this page is a price list. The numbers are illustrative. What matters is the shape of the model and whether it holds up under questioning, which is the point of the conversation this page exists to support.

Known size · the default
$19/mo
illustrative · a size, not a plan

You buy a size and you know what it costs. Use a fraction of it or all of it, the bill is the same. Nobody wants to pay one dollar in January and forty in February for a bakery website.

Your own domain, pointed hereDNS handled for you Website and appSecurity and SSL Served fast everywhereBackups of everything Head start setupsThe assistant A human to callA bigger size when you need it
Measured · for the people who ask for it

Pay for exactly what runs

Developers and heavy users can switch the same pod to measured pricing and pay for what they actually consume. No new product, no migration, no data move. It is a billing setting, not a platform.

You can switch back. Most businesses never turn it on, and most developers never turn it off.

The part that matters more than the number

You will never be surprised by a bill.

The reason people fear usage pricing is not the cost, it is the ambush. Your assistant watches the pod and tells you in plain language before anything changes, so a bigger month is a decision you made rather than a shock you discovered.

You can also set a ceiling you do not want to cross. Real growth gets a heads up. One unusually good day gets absorbed, because a mention on the news is not a new business, it is a Tuesday.

Your assistant
Heads up. Your pod has been about three times busier since the news segment on Friday.
Nothing is wrong and nothing needs doing. If it stays at this level all month, you would move up a size and next month would be about $34 instead of $19. Want me to tell you again if it holds through next week?

Growth in traffic is growth in the business. Paying a little more at fifty thousand visitors a month is alignment, not punishment. What was punishment was the old model, where getting bigger meant a migration weekend and a new invoice for the same website.


What this page still owes you.

The model above is the part that is decided. The numbers underneath it are not, and pretending otherwise would be the fastest way to lose a serious conversation.

Now modelled

Unit economics and market

The cost floor for one pod is now built from published container rates rather than guessed, and the growth model is interactive: running the numbers. Two findings worth knowing before you read it.

  • Memory, not CPU, is what makes an idle pod expensive. Half a gigabyte held open all month costs more than twice the compute a small site actually burns
  • The assistant is the only cost that grows with the feature people came for, and it is the one number on the page with no source behind it

One position that is already settled: this should not try to be the cheapest option on the market. Competing on price against operators with a structurally lower cost base is a losing game, and it is not what the product is for. That argument is in the open questions alongside the rest of what is unresolved.

Where this is going

The model only works inside a business that already exists.

Abuse handling, identity verification, payment processing, network capacity, peering and out of hours support are the expensive, unglamorous foundations underneath any number on this page. An operator that already runs them has a cost base a startup would spend three years rebuilding.

Read the case for why now, the technical page for what is actually under it, and the open questions for what is not resolved.