AIAIAI.FYI
LIVE

LangSmith adds coding agent reports that tie spend to pull requests, now in public beta

The tool traces Claude Code, Codex, Cursor, GitHub Copilot in VS Code and Deep Agents into one dashboard, and it counts toward regular LangSmith tracing usage.

What happened

LangChain has added a dashboard to LangSmith that shows what AI coding assistants cost. It links that cost to pull requests, which are proposed code changes that teammates review and merge. Teams can then see what they got for the money.

Why it matters to you

Companies spend heavily on coding agents but often cannot say what the spend buys. This gives managers and finance a shared view of cost per unit of finished work.

The takeaway

Companies can now check whether their AI coding spend ends in merged code.

What it means for you

If you build software

You can compare cost per merged PR before and after a model change, and find long threads or skills that drive up spend.

If you run a business

Engineering and finance can look at the same cost figures, with unattributed spend shown beside cost per merged PR.

If you invest

LangChain cites $34B of enterprise AI spend on coding in 2026, so cost and ROI tooling is a growing area.

Get stories like this every morning