LangSmith has launched Custom Apps, a new feature that helps users build custom interfaces around their agent data. These apps can be created with chat prompts or coding templates. They run inside LangSmith and do not require hosting or permission setup.
Teams use LangSmith to debug, evaluate, and improve agents. Default dashboards cover latency, cost, and error rates. But many teams want custom views that match their review criteria.
Some teams build their own frontends with APIs and internal tools. These need ongoing maintenance, hosting, and permissions. Custom Apps eliminate this work by providing ready-to-use interfaces.
With Custom Apps, users can share apps inside their workspace. They can reuse workflows for repeated reviews and share results with teammates. This makes reviews faster and more consistent.
Custom Apps are useful for human feedback, experiment reviews, and trace analysis. They can show relevant data slices, compare outputs, and guide reviewers.
Plus plans include one Custom App per organization. Enterprise plans offer unlimited apps. Users can try Custom Apps by signing into LangSmith and reading the docs.
Why it matters
Custom Apps help teams review and improve AI systems more efficiently, saving time and reducing manual work.
What to do
Try creating a Custom App today by logging into LangSmith. Use templates or chat prompts to build your interface.
Source: https://www.langchain.com/blog/langsmith-custom-apps
