Olly FAQs
General
Is Olly available for all Coralogix customers?
Yes. Every Coralogix customer can use Olly.
What is the difference between Olly and Coralogix MCP Server?
MCP is access to Coralogix data. Olly is intelligence. While MCP lets AI tools read observability data, Olly uses that data to actually run production investigations.
What Coralogix MCP is:
- MCP is a way to expose Coralogix API to AI systems like Cursor
- Coralogix MCP exposes logs, metrics, traces, alerts
- It is an integration layer, not a product
What Olly is:
Olly sits on top of the Coralogix Observability Platform and turns data into action. It is an Agentic Observability System made of: Knowledge (understands your services, history, incidents, deployments), Multi-agents (specialized agents for logs, traces, metrics, correlation, security, engineering), Autonomy (autonomously decides how to navigate and solve any challenge), and purpose-built UX for observability use cases.
How accurate is Olly?
Accuracy is one of Olly's strongest advantages. In practice, Olly consistently delivers higher-quality answers than any other SRE Agent, and is one of the few solutions that actually delivers on the promise of autonomous observability. Customers experience this as faster, more precise root-cause analysis, fewer hallucinations, and answers that are grounded in real production context rather than generic LLM responses.
That quality is not accidental. Olly is built around the Context Engineering Triangle: a purpose-built agent architecture with multiple specialized agents, an advanced proprietary knowledge system that deeply understands customer telemetry and environments, and a rigorous agent evaluation system that continuously measures and improves output quality.
What happens if Olly is incorrect?
Like any AI-powered system, Olly is not perfect and may occasionally produce incorrect insights, incomplete analyses, or sub-optimal recommendations. It is built with this reality in mind and is designed to support human decision-making rather than silently replace it. Depending on the task and risk level, Olly operates with appropriate guardrails — ranging from advisory suggestions to actions that require human approval and can be reviewed or reversed.
From both a product and legal standpoint, this possibility is explicitly accounted for. The platform includes transparency, control, and feedback mechanisms to reduce the impact of errors and continuously improve over time, while the legal terms acknowledge the probabilistic nature of AI systems and define clear boundaries and responsibilities accordingly.
On what data tiers does Olly operate?
Olly operates on all Coralogix data tiers — High, Medium, and Low:
- High priority data: Business-critical telemetry fully indexed for real-time querying
- Medium priority data: Monitoring telemetry useful for dashboards, alerts, and trends
- Low priority data: Compliance or archival telemetry with limited query capabilities
Olly does not operate on telemetry in the Block tier, as it is dropped and not stored.
Where does Olly query data from?
Olly currently accesses data from the team's archive. It does not query data from frequent search.
Features
What is the difference between Fast mode and Focus mode?
Fast mode is best for quick, scoped questions like "show the last 50 errors." Focus mode uses a heavier model and is designed for deeper investigations and unknown-unknown scenarios, such as RCA. It may take longer, but provides more thorough analysis.
Can I download or export Olly results along with the queries that were executed?
Yes. Olly allows you to download the results of the queries it executed directly from the UI.
- For artifacts (such as tables), you can use the Export button to download the results as CSV.
- You can also export results from chats and artifacts, including the data Olly generated based on the executed queries.
This makes it easy to share outputs outside of Olly without requiring recipients to have Olly access.
Can Olly preserve context or learn from previous investigations?
Today, the best way to "teach" Olly about your environment, conventions, or known patterns is by using User Rules.
User rules let you:
- Add guidance specific to your systems or architecture
- Capture known behaviors, assumptions, or investigation preferences
- Customize how Olly approaches analysis in your environment
This helps Olly produce more relevant and consistent answers.
Do I need Coralogix access to view shared Olly links?
Yes. Olly supports two sharing options:
- Private (default): Only you can access the link
- Share with team: The link is accessible to members of your Coralogix team
To open an organization-level shared link, the viewer must have access to Coralogix and be a member of the same team. If you need to share results with users who do not have access, exporting the results (for example, as CSV) is the recommended approach.