Obsy AI¶
Obsy AI is the umbrella term for the platform's AI capabilities, including Obsy Copilot for instant personal assistance and Obsy Agent Teams for sustained team work. The two use different ways of working and are suited to different types of problems.
Before use, the workspace owner must enable the workspace AI features in Management > Workspace Settings > AI Analysis. Free features also require this switch to be enabled first.
Product Components¶
| Product | Positioning | Usage | Use Cases |
|---|---|---|---|
| Obsy Copilot | Provided with the workspace; no self-deployment required; the owner must first enable workspace AI features. | Can be opened on any page to ask questions in natural language; on supported pages, it can also perform analysis based on the current object, time range, and filter conditions. | Instant Q&A, anomaly analysis, query statement generation, dashboard and monitor draft generation, Pipeline creation, and note and Runbook capture. |
| Obsy Agent Teams | An Agent workbench that can be configured, deployed, and orchestrated. | Create Agents with clear responsibilities, configure Skills, MCP services, and Channels for them, then trigger them through tasks, automated tasks, or task integrations. | Team division of responsibilities, continuously running intelligent workflows, external system task processing, and result distribution. |
Obsy Copilot¶
Obsy Copilot is provided with the workspace and does not require self-deployment; the owner must first enable workspace AI features. After opening Obsy Copilot, you can directly describe your goal; on supported pages, you can also have it complete tasks using accessible data and capabilities within existing permissions, based on the current object, time range, and filter conditions.
Use it in the following situations:
- You want to quickly understand the current logs, traces, alerts, RUM data, dashboards, or infrastructure objects;
- You need to write or check DQL or PromQL, or filter Explorer data using natural language;
- You need to create dashboards, monitor drafts, or Pipeline parsing rules;
- You want to organize the analysis process into notes or Runbooks, or query product documentation.
Obsy Copilot is designed for instant tasks proactively initiated by users, helping users query, analyze, and create content in the workspace more efficiently and accurately.
Obsy Agent Teams¶
Obsy Agent Teams is used to build and run Agents with clear responsibilities so that they can combine workspace data, external capabilities, and runtime environments to continuously participate in the team's actual work. Each Agent can have independent responsibilities, behavioral boundaries, runtime environments, and available tools.
It is suitable for undertaking complex tasks that require continuous context, multi-step collaboration, and actual operations, for example:
- Organize problem analysis, solution design, operation execution, and result verification into continuous tasks, and maintain context across multiple rounds of processing and cross-tool collaboration;
- Complete analysis based on observability data in the workspace, and within the authorized scope, invoke code repositories, clusters, and engineering tools to directly perform SRE, R&D, or testing work such as code fixes, scaling, and configuration adjustments;
- Connect business systems, external services, or other AI workflows through task integrations, MCP services, and other methods, and bring tasks from different sources into a unified execution pipeline;
- Configure Agents with clear responsibilities for different teams or business scenarios, continuously execute tasks according to schedules, workspace events, or external requests, and send results to configured Channel conversations.
An Agent's actual capabilities depend on the enabled configuration, authorization scope, and deployment environment. After creation, you can view the execution process and results from task conversations and run records.
In workspaces that support this capability, you can also select an Agent directly in the AI drawer to start a task, optionally attach page context, view progress, and open the original conversation to continue asking questions. For details, see Use Agents in the Workspace. Agents also support integration with custom models provided by the team. Model services and data access are subject to the corresponding configuration and permissions.
Quick Access¶
-
Learn how to use Obsy Copilot and its capabilities for intelligent analysis, data querying, content creation, and note and Runbook capture.
-
Learn how to create and deploy Agents, and use tasks and automation capabilities to keep Agents continuously involved in team work.
For how to open Obsy Copilot and manage conversation history, see Obsy Copilot.
Plans and Credits¶
Obsy Copilot and Obsy Agent Teams share the workspace's AI service plan and credits account. For plan versions, credit validity periods, and deduction order, see Plans and Credits.
AI API Call Security¶
AI APIs now include authentication protection based on DPoP (Demonstrating Proof-of-Possession). By binding access credentials to client keys and adding request anti-replay validation, it reduces the risk of misuse after a token is leaked. Calls still follow the permission scope of the account and workspace; continue to safeguard credentials properly.