AI Copilot Metrics

API endpoints for AI copilot usage, adoption, and productivity metrics

AI Copilot Metrics

AI Copilot metrics measure how developers interact with AI coding assistants, adoption rates, and the impact on productivity across your organization. They read from the ai_copilot table: one row per user, per day, per connected copilot source (Cursor, GitHub Copilot, and so on).

These are the seat-level usage metrics reported by the copilot vendors. For line-level attribution captured on engineers’ machines by posthook (sessions, prompts, committed AI lines), see Posthook Metrics.

Available Metrics

Metrics marked multi only can be requested through get-multiple-metrics but not as a single metric_type.

Adoption & Usage

ai-adoption-rate · ratio → ai_adoption_rate

Share of users with at least one accepted line (0–1)

adopted-users · count → adopted_users_count

Number of distinct active users

adoption-rate · ratio → adoption_rate

Same as ai-adoption-rate. Multi only

Suggestions & Acceptance

suggested-lines · aggregation → {agg}_suggested_lines_added, {agg}_suggested_lines_deleted, {agg}_net_suggested_lines_added

Lines of code suggested by AI. Multi only

accepted-lines · aggregation → {agg}_accepted_lines_added, {agg}_accepted_lines_deleted, {agg}_net_accepted_lines_added

Lines of code accepted from AI suggestions. Multi only

ai-acceptance-rate · ratio → ai_acceptance_rate

Accepts ÷ (accepts + rejects), 0–1

acceptance-rate · ratio → acceptance_rate

Same as ai-acceptance-rate. Multi only

total-accepts · aggregation → {agg}_total_accepts

Total number of accepted AI suggestions. Multi only

total-rejects · aggregation → {agg}_total_rejects

Total number of rejected AI suggestions. Multi only

Code Volume

ai-total-loc · aggregation → ai_total_loc

Total lines of code written with AI assistance

ai-percent-ai-generated-code · ratio → ai_percent_ai_generated_code

Accepted lines ÷ all suggested lines, 0–1

ai-code-ratio · ratio → ai_code_ratio

AI lines of code ÷ total lines changed, 0–1

ai-output-estimate · aggregation → {agg}_ai_output_estimate

Daily output estimate attributed to AI-assisted work

total-lines-added · aggregation → {agg}_total_lines_added

Total lines added with AI assistance. Multi only

total-lines-deleted · aggregation → {agg}_total_lines_deleted

Total lines deleted with AI assistance. Multi only

Requests by Type

chat-requests · aggregation → {agg}_chat_requests

Number of AI chat requests. Multi only

composer-requests · aggregation → {agg}_composer_requests

Number of AI composer requests. Multi only

agent-requests · aggregation → {agg}_agent_requests

Number of AI agent requests. Multi only

cmd-k-usages · aggregation → {agg}_cmd_k_usages

Number of inline (Cmd+K) AI usages. Multi only

Cost & Output

ai-pr-count · aggregation → {agg}_pr_count

Number of pull requests created with AI assistance

ai-total-token-usage · aggregation → {agg}_total_token_usage

Total AI token usage

ai-total-cost · aggregation → {agg}_total_cost

Total estimated cost of AI usage

Insights

ai-favourite-model · string → favourite_model

Most used AI model

most-used-programming-language · string → most_used_programming_language

Most used file extension in AI applies. Multi only

Accepted vs Suggested Lines

Retrieve both accepted and suggested lines in a single request to calculate acceptance rate over time.

Request

curl -X POST 'https://api.bilanc.co/metrics/get-multiple-metrics' \
  -H 'Authorization: YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "filters": {
      "team": ["team123"],
      "start_date": "2024-01-01",
      "end_date": "2024-03-31"
    },
    "date_level": "month",
    "metrics": ["suggested-lines", "accepted-lines"],
    "date_fields": {
      "suggested-lines": "date",
      "accepted-lines": "date"
    },
    "aggregations": {
      "suggested-lines": "SUM",
      "accepted-lines": "SUM"
    }
  }'

AI Adoption Rate

Track the percentage of active developers using AI tools over time.

Request

curl -X POST 'https://api.bilanc.co/metrics/ai-adoption-rate' \
  -H 'Authorization: YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "filters": {
      "start_date": "2024-01-01",
      "end_date": "2024-03-31"
    },
    "group_by": ["team_name"],
    "date_level": "month",
    "date_field": "date"
  }'

Most Used AI Model

Retrieve the most commonly used AI model across the organization.

Request

curl -X POST 'https://api.bilanc.co/metrics/ai-favourite-model' \
  -H 'Authorization: YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "filters": {
      "start_date": "2024-01-01",
      "end_date": "2024-03-31"
    },
    "group_by": ["team_name"],
    "date_field": "date"
  }'

Parameter Notes

date_field · string

For AI copilot metrics the only date field is date, the day of the usage record.

aggregation · string

• For line counts, request counts, tokens, cost and output estimate: SUM (or AVG for per-user-day averages)
• For rate metrics (ai-adoption-rate, ai-acceptance-rate, acceptance-rate, adoption-rate, ai-code-ratio, ai-percent-ai-generated-code) and for adopted-users, ai-favourite-model, most-used-programming-language: none

group_by · string[]

Useful columns: source (copilot vendor), most_used_model, model_provider, name, user_email, is_active, client_version, plus the derived fields team_name, squad_name, manager_name. Full list on the overview.

filters · object

Applicable filters for AI copilot metrics:

• ai_copilot_sources: Filter by AI copilot source name (e.g. cursor, github-copilot)
• team: Filter by team IDs
• squad_ids: Filter by squad IDs
• squad_levels: Filter by squad hierarchy level
• manager_ids: Filter by manager user IDs
• departments: Filter by department names
• member: Filter by user merged user IDs
• locations: Filter by employee location
• levels: Filter by employee level

Common Use Cases

• Adoption Monitoring: Track AI tool adoption across teams over time
• ROI Analysis: Compare ai-total-cost against productivity gains
• Acceptance Quality: Monitor acceptance-rate to gauge suggestion relevance
• Language Insights: Use most-used-programming-language to focus AI training
• Usage Breakdown: Compare chat-requests, composer-requests, and agent-requests to understand how developers use AI
• Attribution: Pair with Posthook Metrics to see how much of that usage reaches committed code