A fair way to count multi-step automations
Why Bondi exposes Automation Runs while retaining action-level metering for per-item work, pagination, retries, and AI calls.
An automation that updates one record and an automation that processes 3,000 orders are not the same unit of work. But a price based on every low-level node is difficult to predict.
Bondi therefore keeps two related measurements:
- A billable Action records work that actually happened.
- A public Automation Run groups up to 20 Actions:
max(1, ceil(Actions / 20)).
The trigger and skipped nodes are not Actions. Logic, transformations, waits, and forms count once when invoked. Database, API, and integration work counts per real invocation, including each item and pagination page. AI calls count as Actions and also consume funded AI credit or the Wallet.
Why retain the action count?
Action-level data explains cost. It shows whether a workflow grew because of volume, pagination, retries, or model calls. It also makes an apparently small workflow comparable to a batch process.
Why publish Runs?
Runs make capacity understandable. Standard includes 2,500 per month and Pro includes 15,000. A 5,000 Run capacity pack costs $49 per month. After that capacity is used, each additional Production Run costs $0.02 and is deducted automatically from the Wallet. The pack remains the more predictable option for steady usage.
The goal is predictability without pretending all executions perform the same amount of work.
Build your first system
Start a 30-day trial. No card required.