By design, our service is autonomous and it’s not uncommon for us to take 1,000s of optimization actions a month for a reasonably sized, dynamic environment (sometimes at 5am on a Saturday). At that scale, inline reviews, approvals, vetoes, etc. aren’t feasible and it’s extremely difficult for humans to evaluate individual transactions without the full context of what our algorithms are seeking to holistically accomplish (i.e., an individual transaction may look nonsensical).
Obviously, customers need to be able to control and direct the activities of our service, so instead of inline, per transaction approvals, we frontload how you govern our service via settings. The settings allow you to define how much base commitment you want, how aggressive/conservative our flex coverage is vis a vis how much of a decline in compute usage you want to be able to absorb, etc.
Our service’s job is then to monitor your environment in real-time and handle the hour to hour activities of maximizing your savings and minimizing your commitment risk, within those setting constraints. You control settings and can change them at any time.