Test Explorer

Test Explorer is an interactive workspace for building a test set. Draft or import tests, invoke your endpoint, and see responses and metric scores in the same place — so you can iterate until you find the gaps you care about.

Two benefits of Test Explorer

Test Explorer surfaces two complementary capabilities:

1. Interactive endpoint play

Write a test, invoke the endpoint, see metric scores — all in the same view. No test run setup and no waiting for a batch job. The loop is fast enough to tweak a single prompt and confirm the fix in seconds.

2. Exploration via suggestions

Given the tests you already have, Test Explorer samples up to 10 of them as examples and asks an LLM for 20 new inputs. Each suggestion is embedded; the batch is re-ranked by diversity so the prompts that differ most from the other suggestions in the same run appear first. The endpoint and your metrics run in parallel as results stream in. You read the ranked table, then accept only the rows that add real coverage or expose real failures.

For the full pipeline (topic-scoped runs, generation guide, progress bar, accept flows), see Building and Evaluating — Suggestions and Workflow.

What you can do

• Metrics at a glance — each test and topic shows pass/fail and per-metric scores
• Organize by topic — group findings (for example off-topic or competitor cases) for export and regression runs

Prerequisites

• At least one endpoint configured in your organization
• At least one metric available under Metrics

Get started

Open Testing → Test Explorer in the sidebar.

• Click New session to start from scratch
• Click Load Test Set to seed an explorer session from an existing test set

Pages in this section