Skip to main content
Design Preview

Value Telemetry

Understand what your agentic spend delivers.

Link your AI activity to the work it actually delivered, issue by issue. CodeLantern provides the metrics to understand what the work required, including actual cost for GitHub Copilot customers, so you can plan ahead and account for what’s behind you.

The Challenge

Agent activity is growing. The work-level context is missing.

Provider dashboards report activity and, in GitHub Copilot’s case, billed usage. They do not connect that activity to the issue, its size, or the work record your team needs to interpret what happened.

Usage totals, without work context

Provider reports show activity or spend, but not the issue, size, or decisions behind it.

No link to the issue

Nothing connects agent activity to the work item where the team can interpret it.

Design Preview

See cost and effort at the level of the work.

Value Telemetry is available for evaluation with prospective design partners. It connects provider-appropriate session signals to real issues, so you can understand agent activity at the level where work is planned, reviewed, and delivered.

Cost by issueGitHub Copilot billed usage
Q3 · Platform hardening5 issues
IssueCost
Add OAuth device-code flow$41
Fix pagination off-by-one$192
Refactor billing webhooks$88
Add CSV export to reports$23
Migrate to Postgres 16$67
Total$411

GitHub Copilot cost, tied to the issue

Accurate GitHub Copilot billed usage is attached to the issue, its size, and its sessions, so you can understand where that spend accumulated.

Sessions, grouped by the work

Open a session to see duration, model time, token volume, skills, MCP calls, and its OTEL timeline, so you can understand how that session contributed to the issue.

Down to the session

Open a session to see its duration, model time, token mix, skills, tools, and timeline, so you can investigate how the effort accumulated.

The work record beside the telemetry

Issue size, contributors, solution design, decisions, and session history stay connected, so you can interpret activity in the context of the work.

Coming Soon

Find unusually intensive work sooner.

We’re developing a normalized agentic-effort score and anomaly detection across GitHub Copilot, Claude Code, and Codex, so you can identify and investigate issues requiring unexpectedly high effort relative to similarly sized work.

What it enables

Forecast GitHub Copilot spend.

With issue-specific GitHub Copilot cost tracking, you can build a history of comparable completed work and forecast the likely cost of future issues, so you can budget agentic development with greater confidence.

The Method

Structure is what makes value measurable.

A provider console can’t tie activity to the full work record: it sees usage, not the design and decisions. Our methodology organizes agentic work around the unit you already manage: the issue.

Small, shippable issues the way strong teams already work: agentic agile vs. spec-driven waterfall and software factories that fail. That’s exactly what makes cost and effort legible, issue by issue.

Every issue runs the same lane, and each one enriches a perpetual knowledge base that travels with your project, so the next issue starts with more of the team’s context already available.

Content-free by design

See the shape of the work, not its contents.

Value Telemetry records operational signals from a session: timing, model activity, token use, skills, and MCP tool calls. It does not collect prompts, transcripts, source code, or diffs, keeping the evidence focused on how work moved through the system.

Official Partner

Get started

See it in action.

Value Telemetry is a Design Preview. We’ll show you how the current issue and session signals apply to work you’re already shipping and discuss whether an evaluation is a fit.

  • A short intro call to understand your team
  • We show it on your real work, not a canned demo
  • An honest go / no-go, no pressure
Optional