One path, from a drained node to a number on the invoice.
Eight steps from your scheduler’s logs to a number on the invoice.
One path, from a drained node to a number on the invoice.
A node goes down, gets drained, or starts running sick. Your scheduler already knows.
A node goes down, gets drained, or starts running sick. Your scheduler already knows.
Counting it yourself is the hard part.
- Gone in an hour
- Does not count
- Window closed
- Kept, every month
Illustrative.
One path, four pairs.
Every hour gets one label.
- Counts against the provider
- Excluded
- Delivered
Illustrative.
The rules come from your order form, parsed once and confirmed by a person, then applied deterministically. Never an LLM in the money path, so the number holds up when the provider replies.
Ask it what you got.
- Counted by rules
- Explained by an agent
- Sent by you
Illustrative.
100%
of node-hours labelled, the pilot’s first success criterion
1 week
from export to statement in hand
100%
of node-hours labelled, the pilot’s first success criterion
3
possible verdicts per hour, each with the clause cited
0
changes to your cluster; the collector only reads
Questions
Because the provider’s own monitoring decides what counts as down, and a node that stays up but runs sick counts as fine. Credits are capped and expire within weeks unless you file with evidence. The statement applies the provider’s definitions first and your order form second, so the number survives the reply.
What percent of the nodes you paid for did you actually get last month?
Send one export and find out. No install, no call first, no product to evaluate.
Get the free statement