Features overview

Once telemetry is flowing, the dashboard turns your spans into these areas:

Area	What it's for
Summary	KPIs, token charts, top services and models at a glance
Traces · in development	Explore every trace and span; a live stream; the swimlane flow view
Reporting & Usage	Spend over time, per-model/service breakdowns, AI-written reports, schedules
Alerts	Rules on error rate, latency, and token/cost budgets, with a one-click jump to the failing traces
Detect Issues	LLM analysis of traces across quality categories
Evaluation · in development	Datasets, LLM judges, and evaluation runs
Agent groups	Drag-and-drop board mapping agents to business areas (Finance, Sales, …)
MCP server	Query your telemetry in natural language from Claude Desktop or any MCP client
AI Asset Dashboard	Catalogue of approved agents, MCP servers and skills: publish, discover, install

How the areas fit together

The same ingested spans feed every area — observability on the left, evaluation on the right, with SAP AI Core powering the LLM-assisted parts (judges, AI-written reports, Detect Issues) and the pricing/model-lifecycle data.

Dashed lines: features that call SAP AI Core — Reporting's AI-written summaries, Detect Issues, and the Evaluation judges. Alerts and cost views also read AI Core's foundation-model catalog for prices and retirement dates.

New here? Start with Installation and Connect AI assets — the features come alive once real spans arrive.