Anthropic: Claude writes 90% of its team's code
How the Claude Code team verifies agent work
Coding tools
Agent spend tied to the pull requests it produced
Written with AI assistance, reviewed by the ACN desk.
“Every company now wants to know what all that spend returns”
LangChain has added reporting built for coding agents to LangSmith, its platform for tracing AI agents. LangChain’s announcement says the feature ties coding agent spend to outcomes such as pull requests.
A new concept called outcomes links GitHub pull requests or GitLab merge requests to agent threads, according to LangChain. The spend in each trace, a record of one agent run, is attributed to the pull requests that agent worked on. LangSmith splits that spend by pull request state: open, closed or merged. Spend that never produced a pull request is tracked separately as unattributed.
LangChain says the feature supports Claude Code, OpenAI Codex, Cursor and GitHub Copilot in VS Code. It also supports Deep Agents, for teams building a custom coding agent. The feature is in public beta on LangSmith Cloud. Coding agent traces count toward regular LangSmith tracing usage, the company says.
LangChain cites Menlo Ventures in saying coding is the largest category of enterprise AI application spend, at $34 billion in 2026. The company says most engineering leaders still cannot say exactly what that spend is buying. LangChain also says spending on its own coding agent, Robin, started climbing.
“Every company now wants to know what all that spend returns”
“the spend in each trace is automatically attributed to pull requests each agent contributed to”
Sponsor
Spot an error? Email corrections@aicodingnews.si. Fixes are logged on the corrections page.
Sponsor
Sponsor
How the Claude Code team verifies agent work
Code moves to Verse · more assets stored as text for AI
New default for agent sessions · Local harness removed Feb 2027
Adaptive effort · secret detection · public preview
Sponsor
Sponsor