Use AI to Discover
What to Measure
Watch SLO Discovery in Action
See how a guided interview moves from a plain service description to structured SLO proposals.
Most SLO Programs Stall Before the First SLO Is Written
Blank-slate paralysis
Teams stall on the very first design decisions, long before any configuration begins. Your monitoring platform cannot tell you what matters to your customers until they start complaining.
The onboarding gap
People new to SLOs have to learn the concepts before they can contribute to SLO design. That learning curve slows adoption across the whole organization.
From signal to reliability
Raw telemetry data does not tell you if your users expect perfection, knowing what "good enough" looks like can save you money and time.
Upstream of every SLO tool
Every SLO platform assumes you already know what to measure. SLO Discovery works on the decision that happens before a single SLO gets created.
How a Discovery Session Runs
Describe your service
Give a short, concrete description of your service and what it does. That seeds the session with enough context for the assistant to ask useful questions.
Guided interview
The assistant asks targeted follow-up questions on system boundaries, critical user journeys, failure modes, and acceptable quality thresholds. A topic panel tracks coverage so your knowledge gaps stay visible as you go.
Generate SLO definitions
Once there is enough context, choose Generate SLO definitions to turn the conversation into structured proposals that Nobl9 can evaluate.
Review proposals
Proposals land in the SLO proposals panel. Your team reviews, refines, and decides which ones carry forward into Nobl9's full SLO creation flow.
Structured Proposals With the Context Your Team Needs
Aspect being measured
Each proposal names the specific reliability dimension under consideration, so there is no ambiguity about what the SLO would track.
Success indicator
The signal or metric that would confirm the SLO is being met, grounded in the context you gave during the interview.
Suggested target with rationale
A draft reliability target and an explanation of why that threshold was proposed, so your team has something concrete to debate.
Failure risks
What could push this SLO into a breach, based on the failure modes and quality thresholds you described in the session.
Error budget implications
How the proposed target translates into an error budget, so teams can weigh reliability investment against engineering velocity before committing.
Session management and feedback
Track, rename, and revisit past discovery sessions. An explicit feedback mechanism lets your team help improve the underlying model over time.
Built to Support Your SRE Team's Judgment
Non-destructive by design:
Nothing is created in your organization during a discovery session. Every output arrives as a draft proposal, and your team stays in control of what gets built.
Human-in-the-loop review
Every proposal requires explicit review and refinement in the SLO proposals panel before anything is acted on. The assistant suggests, your team makes the call.
Tackles the upstream problem
SLO Discovery works on the "what should we even measure" question that comes before Nobl9's guided or policy-based SLO creation flows. It fills the gap most SLO tooling assumes is already solved.
Structured output for review
The result is draft guidance written for people to read and discuss, giving your team a shared starting point for the SLO design conversation.
Your Monitoring Stack, Your Rules
When your team is ready to act on a proposal, the resulting SLOs live in the Nobl9 platform, which monitors-agnostic and plugs into the tools your organization already uses via native integrations and the SLI Connect ingestion engine.
Before You Start a Session
Still experimental
SLO Discovery is an experimental Nobl9 Labs capability, its behavior and output may change without notice while the feature matures.
Switched off by default
The feature is not enabled out of the box. Reach out to Nobl9 support to turn it on for your organization.
Everything stays a draft
Output is draft guidance for your team to review. No SLOs are created in your organization until your team explicitly acts on a proposal.
Session history may not persist
Chat history may not be durable between sessions, treat the feature as unsuitable for regulated, contractual, or auditable environments.