Olly FAQs
General
Yes. Every Coralogix customer can use Olly.
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.
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.
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.
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.
Olly currently accesses data from the team's archive. It does not query data from frequent search.
Features
Olly previously offered three chat modes — Pro, Focus, and Fast — to balance speed and depth. The current single chat mode runs on the same skill-based architecture, handles both lookups and deep investigations, and outperforms the older Fast and Focus modes for most observability work, so the three-mode picker has been retired from the UI.
The Olly API keeps accepting focus and fast as interaction_mode values for backwards compatibility — both route to the single mode (skill). Migrate existing integrations to skill when convenient.
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.
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.
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.