Dashboard Overview

The dashboard is at app.tandemu.dev . Log in to see your organization’s data.

All dashboard pages support team and time range filters via the controls in the top-right corner.

Dashboard Home

Path: /

The home page is the command center with org-wide KPI cards and four charts.

KPI Cards

• Total Sessions — Claude Code sessions across the team
• AI Code Ratio — Percentage of code generated by AI vs written manually
• Active Developers — Team members who had sessions recently
• Total Lines of Code — Combined output (AI + manual)
• Avg Cycle Time — Average task duration from /morning to /finish
• Tool Success Rate — Percentage of successful tool executions in Claude Code

Charts

• AI vs Manual Code — Donut chart showing the current AI ratio
• Tool Usage — Horizontal bar chart of Claude Code tool calls by type
• Task Velocity — Weekly average task duration trend
• Investment Allocation — Where engineering time goes (features vs bugs vs tech debt vs maintenance)

Activity

Path: /activity

Developer sessions, time tracking, and code-level analysis:

Stats Cards

• Total Active Time — Combined developer session hours
• Total Sessions — Number of Claude Code sessions
• Avg per Developer — Average session time per team member

Components

• Activity Chart — Daily session and hour counts over time
• Session Log — Table with date, developer name, active time, session count
• Developer Leaderboard — Per-developer table with sessions, active time, AI lines, manual lines
• Hot Files — Most-changed files ranked by commit frequency
• AI Effectiveness — AI-written lines by file (where AI output survives)

No manual timesheets needed. Time is tracked automatically from Claude Code sessions.

Insights

Path: /insights

AI investment analysis for engineering leads.

Stats Cards

• AI Lines — Total AI-generated lines of code
• Manual Lines — Total manually-written lines
• Tasks Completed — Number of finished tasks
• Total AI Cost — Estimated AI token spend

ROI Metrics

• Productivity Multiplier — How much more output the team produces with AI assistance compared to manual-only coding
• Capacity Freed — Hours of developer time freed by AI assistance per period
• Cost per Task — Average engineering cost per completed task

An assumptions banner at the top lets you configure the developer hourly rate used for cost calculations.

Tandemu Impact

• Memory Hits — How often Claude accessed stored memories during sessions
• Friction Trend — Whether friction is increasing or decreasing over time
• Knowledge Shared — Organization memories created and published

Charts

• Throughput Chart — Tasks completed over time, showing velocity trends
• Cost Efficiency Chart — Cost per task over time
• Token Usage — AI token consumption patterns across the team
• AI Adoption Leaderboard — Per-developer ranking by AI code ratio

Friction Map

Path: /friction-map

Identifies files and components where developers get stuck:

• Repository paths ranked by friction score
• Prompt loop count — How many times developers re-prompted Claude on the same issue
• Error count — Tool execution failures
• Sessions count — Number of unique sessions affected per path
• Severity badges — Based on a weighted score: promptLoops + (errors × 2). Low (below 10), Medium (10-19), High (20+)

High friction on a file usually means:

• The code is complex or poorly documented
• There are subtle bugs that AI can’t easily fix
• The architecture needs refactoring

Use this to prioritize tech debt cleanup — focus on the files that slow your team down the most.

AI Memory

Path: /memory

Browse and manage your team’s AI knowledge base. See Memory Insights for the full guide.

Stats

Four KPI cards: Total Memories, Personal, Organization, Memory Health (% actively used by Claude).

Charts

• Memory Categories — Bar chart showing knowledge distribution (architecture, patterns, gotchas, decisions, etc.)
• Memory Coverage — Donut showing what percentage of memories Claude actually references

Insights

• Knowledge Gaps — Modules with frequent changes but few/no documented memories
• Most Referenced — The memories Claude relies on most
• Cleanup Candidates — Unused memories that may be stale or irrelevant

Browser

Search, filter, edit, and delete memories. Toggle between Personal and Organization scope. View as a list grouped by repo or browse as a file tree.

DORA Metrics

DORA metrics are calculated from merged GitHub PRs, synced automatically every 4 hours.

The dashboard shows a DORA card with performance rating badges and a weekly trend chart. You can trigger a manual sync via POST /api/telemetry/github-sync.