LangSmith Launches Custom Apps for Building Tailored Interfaces on Agent Data
Key point
LangSmith has launched Custom Apps, allowing Plus and Enterprise users to build and publish custom interfaces on agent data, with Plus plans including one app and chat-based creation.
Details
LangSmith has introduced Custom Apps, a feature enabling developers to build, publish, and run custom user interfaces directly on top of LangSmith data. This addresses the need for teams to interact with agent telemetry in ways that align with specific quality criteria and review processes, moving beyond prebuilt dashboards for latency and error rates.
How Custom Apps Work Users can create interfaces via a chat prompt or by coding directly. On Plus and Enterprise plans, users can instruct LangSmith Chat to generate a working app based on desired data visualization. Alternatively, developers can use provided templates and coding agents to build against the LangSmith API. Once published, these apps become shared interfaces within the workspace, eliminating the need for separate hosting, authentication, and permission management.
Key Use Cases
- Tailored Annotation Interfaces: Teams can create review UIs specific to reviewer roles, such as showing full traces to engineers while hiding technical details from subject matter experts.
- Customized Experiment Reviews: Users can build interfaces to compare outputs across prompt versions or model swaps, slicing results by customer segment or failure category for release reviews.
- Focused Trace Analysis: Teams can create UIs around specific recurring questions, such as identifying where behavior changed or which step caused a failure.
Availability and Pricing Custom Apps are available on Plus and Enterprise plans. The Plus plan includes one Custom App and access to the chat-based creation feature. Enterprise plans offer unlimited Custom Apps. Users can access the feature via the LangSmith platform and documentation.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.