-1.png)
Where Does Your Team Stand
Seven questions, answerable in one meeting. Each "no" indicates an area where the asymmetry is currently unmanaged.
Read the full report: The Velocity Asymmetry →Can you state, in machine-readable form, what "reliable enough" means for your five most critical services?
In most organizations, there is no single place where an engineer or an agent can look up what counts as acceptable.
When a service degrades at 2 a.m., does the on-call engineer see a single healthy-or-not verdict in seconds, before consulting a dashboard?
Dashboards locate the symptom. The model in someone's head supplies what it means.
Do release and rollback decisions rely on error budget as a matter of routine?
A burning budget signals that changes deserve more scrutiny, from people and from agents alike.
Are SLOs defined for a new service in the same pull request that moves it to production?
Services accumulate faster than anyone remembers to watch them.
Has every objective been reviewed since the service it describes last materially changed?
A target accurate for a year at human pace can lose its meaning in a week when agents extend the service daily.
Does every objective have a named owner who would notice if it went stale?
A stale target still looks like coverage on a status page.
Can an AI agent query reliability state before acting on production, through an interface built for programs?
An AI reviewer without a written target inherits the same blindness as an AI writer.
Partials suggest you are closer than the score reads.
Email me my results and what to fix first
Your score, how it compares with other teams, and a short recommendation for each "no" or "partially".
The form could not load (this can happen with strict content blockers). You can still grab the report directly below.
The seven questions, in full
The seven questions below can be answered in a single meeting. Each "no" indicates an area where the asymmetry is currently unmanaged.
- Definition. Can you state, in machine-readable form, what "reliable enough" means for your five most critical services?
- Verdict. When a service degrades at 2 a.m., does the on-call engineer see a single healthy-or-not verdict in seconds, before consulting a dashboard?
- Budget. Do release and rollback decisions rely on error budget as a matter of routine?
- Coverage. Are SLOs defined for a new service in the same pull request that moves it to production?
- Time. Has every objective been reviewed since the service it describes last materially changed?
- Ownership. Does every objective have a named owner who would notice if it went stale?
- Machine access. Can an AI agent query reliability state before acting on production, through an interface built for programs?
Scoring
Zero to two yes answers means reliability is reactive and incidents are the discovery mechanism.
Three to five means the contract exists in places, and drift and coverage gaps are the risk.
Six or seven means judgment scales with the system and autonomy can widen safely.
From The Velocity Asymmetry, the Nobl9 report on AI-era reliability.