Interactive Demo

The XeroML demo is a live, shared project that gives you view-only access to explore the platform’s features before connecting your own application. Real users interact with the demo chatbot, so you’ll see genuine traces, sessions, scores, and feedback — not synthetic data.

Tip

You can explore the demo without signing up. Create a free account to also access the example prompts, datasets, and experiment runs.

What’s in the Demo

The demo project contains traces generated by users interacting with the example chatbot. Each trace represents a real LLM interaction and is fully instrumented with:

• Structured observations — every LLM call, tool use, and retrieval step
• Session grouping — multi-turn conversations linked across traces
• Evaluation scores — automated LLM-as-a-judge scores and human feedback
• Prompt versions — the prompts used are managed in XeroML and linked to each trace
• Cost and latency data — token usage and timing for every operation

Getting Started with the Demo

1. Access the demo project

   Visit XeroML Cloud and open the example project from your dashboard. No credit card required.

2. Browse traces

   The Traces page shows every chatbot interaction as a row. You can see timing, costs, input/output summaries, and evaluation scores at a glance. Click any trace to inspect the full execution graph.

3. Explore a trace in detail

   Inside a trace you’ll find:

   • The nested observation tree (spans, generations, events)
   • The exact prompt and completion for every LLM call
   • Token usage and cost breakdown
   • Any scores applied by evaluators or human annotators

4. Review sessions

   Navigate to Sessions to see multi-turn conversations. The session replay shows how context evolved across turns in a single view.

5. Check the prompts

   Open the Prompts section to see the versioned prompts powering the demo chatbot. Each prompt is linked to the traces that used it, so you can compare performance across versions.

6. Run an experiment (with account)

   Clone a dataset from the demo project and run your own experiment against it to see how XeroML evaluation workflows operate end to end.

Try It Yourself

The demo chatbot is open source and fully instrumented. All interactions are visible in the shared project, so you can see your own traces appear in real time after chatting.

After exploring the demo, the recommended next steps are:

1. Instrument your own application
2. Move prompts into XeroML
3. Set up systematic evaluations