Skip to main content
LangSmith+17 Mixed

Guides

How do we get more value from LangSmith?

  1. LangSmith Tracing & Evaluation Playbook

    Trace the agent without sensitive fields. Turn failures into datasets and test each prompt or model change before release.

    IncludesInstrumentation, dataset, and evaluator checklist, plus the release gate to agree

    1. Before workInstrument and set retention
    2. During workTrace, annotate, and collect
    3. Before mergeRun the experiment gate
    • Testing
    • Developer Workflow

  2. LangSmith Trace Spend & Retention Checklist

    Set each project's default retention. Add a monthly usage limit and decide which evaluators may upgrade traces to longer retention.

    IncludesRetention and usage-limit worksheet

    • AI Costs
    • Engineering Management

  3. LangSmith Data Handling & Hosting Checklist

    Keep restricted fields out of traces. Choose a region and determine whether the workload needs hybrid, BYOC, or self-hosting.

    IncludesSecurity review checklist

    • Security
    • Governance

  4. LangSmith Deployment: Choosing How Your Agents Run

    Price Serverless and Dedicated in LSUs before launch. Compare Cloud and hybrid with a standalone Agent Server.

    IncludesDeployment option map and monthly cost worksheet

    • Agent Development
    • AI Costs

View all guides