Skip to main content

Analytics

Use Analytics to see how you and your organization use Devs.ai over time: chats and messages, which models run, and where traffic comes from. If you use the coding CLI, you also see sessions, repositories, and change outcomes in the Code view. You can focus on your own usage, or if your role allows it, you can switch to Organization and see activity across your whole org.

📝 Note: Your organization must have the updated analytics experience turned on before you see these screens and filters. If you still see the legacy analytics page, contact your platform administrator.

Opening the dashboard​

  1. Sign in to Devs.ai.
  2. In the sidebar, click Analytics.

You land on the dashboard with filters already set, such as your personal usage for the current month. You can change scope, category, or period at any time, and the charts and tables refresh for the combination you pick.

Choosing who the data is for​

At the top of the dashboard, select Me or Organization:

  • Me: Activity tied to your account. You see your chats, your tokens, and in Code view your CLI sessions.
  • Organization: Activity across members of your organization. You only see this option when your role allows usage reporting across your whole organization.

If you do not see Organization, you still have full access to your own metrics.

Choosing what to measure​

Use the category controls to switch the lens:

  • All: Chats with and without agents.
  • Agents: Chats that use an agent. When this lens is active, you can narrow further to internal users (people in your organization), external users (outside your organization), or all agent traffic.
  • Code: CLI coding sessions from the integrated coding flow, not general web chat. Here you see repositories, edits, acceptance, and token cost for those sessions.

📝 Note: Message style metrics intentionally exclude rows that come from the CLI in the general All and Agents views, so you are not double counting the same work you see under Code.

Choosing a time range​

Pick a preset (Today, Week, Month, Quarter, Year, or All time) or you choose Custom and set a start and end date.

When you pick a preset other than All time or Custom, you can turn Rolling on to use rolling windows instead of calendar boundaries, such as “last 30 days” instead of “this calendar month.”

When your org offers storage analytics in Files, those numbers use running totals, not an hourly time series. In that case you usually stay on All time and other range options may be disabled.

What you see​

You get cards and charts that load together for the filters you chose. What you see depends on whether you picked Me or Organization, and whether you are in All, Agents, or Code, but you generally see the following.

Summary row​

  • Message views: View highlights such as total chats, total tokens (with message counts), average tokens per user in organization views, and active users in organization views. In Me view, you may see context about your subscription when your plan exposes it.
  • Code views: View highlights such as changes (individual code change events), tokens, files changed (with line totals), acceptance rate, and estimated token cost with a breakdown of input and output.

Charts​

Charts provide data on token usage over time. In Code view, token usage and cost often appear on the same chart. You also see charts that break down token usage in more detail, model distribution in message views, and source distribution (where conversations start from) in organization message views.

In Code view, a code analytics chart summarizes outcomes such as accepted, auto accepted, and denied changes.

Top lists and browse​

  • In the Organization view you get top users and sortable, searchable browse lists so you can scroll or search past the first page.
  • In the Message view you can see data on top agents.
  • In the Code view you see data on top repositories.

Click a row in a top list or browse table to open a detail panel with more stats and, when it applies, nested lists (for example chats for a user, or repositories for a coder).

Detail panels​

When you open a user, agent, or repository from a ranking or browse list, you see a focused summary for that entity. You might see total tokens, chat or session counts, cost when it is available, and in Code view you see lines and files created, edited, or deleted plus acceptance metrics. You use these panels to answer questions like who is driving cost or which repository saw the most activity, without exporting data first.

Permissions​

You need read access to Analytics to open the dashboard at all. To see totals across your organization, you also need permission to read organization usage (or the equivalent reporting access at the organization level). If either check fails for you, you may not see the dashboard or the Organization scope.

Limitations​

  • Rollups: You are always looking at data stored at hourly granularity. When you pick a longer range, you see numbers aggregated from those hourly buckets.
  • Percentiles: Where you see response or RAG timing percentiles, values that combine many hours may be approximate. That is the same limitation you see in most combined percentile views.
  • Feature rollout: The people who administer your environment can turn on ingestion and the UI on different schedules. If numbers look incomplete for a recent day, your organization may still be backfilling history or finishing pipeline work.

If something does not match what you expect, you confirm your category, period, and whether you are on Me or Organization, then you ask your administrator whether analytics is fully enabled for your organization.

⌘J