Telemetry
Telemetry and Analytics
Telemetry and analytics in Peeps AI play a vital role in enhancing system performance, monitoring usage patterns, and driving informed decisions for framework improvements. This feature provides developers with actionable insights into how Peeps AI agents and workflows operate while ensuring robust privacy and data security. By understanding telemetry and analytics, you can optimize your AI systems, improve agent performance, and contribute to the community-driven evolution of Peeps AI.
Goals of Telemetry in Peeps AI
Telemetry in Peeps AI serves the following key purposes:
❖ Performance Optimization: Analyze execution times, agent efficiencies, and task completion rates.
❖ User Insights: Understand which features, tools, and configurations are most frequently used.
❖ Improved Development: Gather anonymous data to guide the prioritization of new features and improvements.
❖ Community Empowerment: Enable users to opt into detailed analytics for deeper insights and contributions.
Data Collected
Peeps AI collects anonymous telemetry to maintain user privacy while gathering valuable insights. The data collected includes:
Framework Usage:
Version of Peeps AI and Python.
Operating system details (e.g., macOS, Linux, Windows).
Number of CPUs and system specifications.
Agent Metrics:
Roles and goals of agents (e.g., Researcher, Analyst).
Number of agents and tasks in workflows.
Whether agents use memory or delegation.
Process Metrics:
Types of workflows (e.g., sequential, hierarchical).
Task execution methods (parallel vs. sequential).
Language Models:
Models used (e.g., GPT-4, LLaMA) and provider preferences.
Token usage statistics (if applicable).
Optional Extended Telemetry
Developers can opt into further telemetry by enabling the share_group setting in their Peeps configurations. With this enabled, the framework collects additional, task-specific data such as:
Agent backstories and goals.
Detailed execution results.
Context and outputs of workflows.
Note: This extended data is shared only with explicit consent and is used to improve the framework while respecting user privacy.
Privacy and Security
Peeps AI ensures robust privacy protection with the following measures:
❖ Anonymized Data: No sensitive information such as API keys, prompts, task descriptions, or responses is collected by default.
❖ User Control: Telemetry can be entirely disabled by setting the environment variable OTEL_SDK_DISABLED=true.
❖ Opt-In Only: Detailed telemetry is collected only when users enable share_group.
❖ Transparent Practices: Telemetry data is never shared with third parties and is solely used for Peeps AI development.
Enabling and Configuring Telemetry
To enable or disable telemetry in Peeps AI, follow these steps:
Disabling Telemetry
Add the following line to your .env file to disable all telemetry:
Enabling Extended Telemetry
To share detailed telemetry data for community-driven insights:
❖ Enable telemetry in the Peeps configuration:
❖ Alternatively, set the share_group attribute in the agents.yaml configuration:
Accessing Analytics in Peeps AI
Peeps AI provides several tools for users to analyze telemetry data in their projects:
Built-in Metrics Dashboard
Peeps AI includes a lightweight analytics dashboard that can be accessed via the CLI:
This dashboard displays:
Agent and task statistics.
Execution times and bottlenecks.
Most commonly used roles and tools.
Exporting Data
Users can export telemetry data to external analytics platforms for further analysis:
Export to JSON:
Export to CSV:
Integration with Third-Party Tools
Peeps AI supports integration with third-party analytics platforms (e.g., Grafana, Kibana) for advanced telemetry visualization. You can configure this in the .env file:
Last updated