Rollout

AI Code Insights scales from a small pilot to organization-wide deployment via MDM. This guide covers rollout strategy and deployment paths.

Start with 5-15 developers who use a mix of AI coding agents (IDE-based and CLI-based). Run the pilot for 2-4 weeks — long enough to capture multiple push cycles and build confidence in the data.

What to evaluate

  • Data accuracy — do attribution percentages match developer self-reports?
  • Agent coverage — are all agents in use detected?
  • Deployment friction — did the installer, EDR, and OS permissions work cleanly?
  • Developer reception — were questions answered by Security and privacy?

Expanding to the organization

Once the pilot validates data quality and deployment:

  1. Pre-configure EDR exclusions for daemon processes and paths (see Endpoint considerations).
  2. Communicate to developers.
  3. Deploy via MDM for the broadest coverage with the least manual effort.

Deployment paths

AI Code Insights is distributed as a signed installer for each platform. Download the latest installers from Admin → AI Code Insights.

For individual installs (evaluation, pilots, or small teams), see the Installation guide.

MDM deployment

For organization-wide rollout. Deploy the daemon via Jamf, Intune, Ansible, or your MDM platform of choice, with pre-provisioned configuration so developers do not need to set anything up manually.

See the MDM installation guide for installer behavior, configuration delivery, and EDR exclusions.

Monitoring your rollout

Use the Client versions panel on Admin → AI Code Insights to track which daemon version each developer is running. See Configuration for details.

Data begins appearing in DX reports after the first push from a monitored repository.

Positioning internally

AI Code Insights measures AI agent adoption and effectiveness. It does not measure individual developer productivity, track time, or evaluate how developers write code.

Tie AI Code Insights to organizational goals that developers share:

  • AI adoption: Understand which agents are being used and where adoption gaps exist, so the organization can invest in the right tools and training.
  • Agent experience: Identify where AI agents hit friction — missing context, ambiguous instructions, codebase barriers — so platform teams can remove obstacles.
  • ROI: Quantify the dollar impact of AI agents to justify continued investment in the agents developers rely on.

Lead with the organizational value and be transparent about what data is collected. See Security and privacy for the full data collection breakdown.

What developers typically ask

Does this read or transmit my source code? No. Only aggregate metrics and commit metadata are sent to DX.

Does this log my keystrokes? No. The daemon observes typing cadence locally to distinguish AI-inserted code from manually typed code.

Does this monitor files outside my work repositories? No. Only repositories whose Git remote origin matches a repository in your organization’s Data Cloud.

Can my manager see my code? No. Managers see aggregate metrics — such as what percentage of a team’s committed code is AI-authored — not individual lines of code.

Is this data used for performance reviews? AI Code Insights measures code authorship at the team and organization level, not individual developer performance. How your organization uses DX data is determined by your organization’s policies.

Can I verify what’s happening on my machine? Yes. Run aicodemetricsd status to see what the daemon is doing. Local databases are unencrypted and inspectable with any SQLite client.

Next steps