# PowerTraces documentation

Set up OpenTelemetry for PowerTraces: endpoints and keys, every language, AI tracing for Claude, OpenAI, Gemini and other providers, and what each page needs.

## Guides

- [What PowerTraces is](https://powertraces.com/docs/overview): The product in one page: signals, pages and how it's sold.
- [Live demo](https://powertraces.com/docs/live_demo): Trying PowerTraces without signing up.
- [Getting started](https://powertraces.com/docs/getting_started): From sign-up to the first trace in about five minutes.
- [Sending telemetry (endpoints, keys, languages, Collector)](https://powertraces.com/docs/sending_telemetry): Endpoints, auth headers, environment variables and per-language setup.
- [Keys and credentials](https://powertraces.com/docs/credentials): How server keys and browser keys work.
- [What each page needs](https://powertraces.com/docs/unlock_pages): Which attributes turn on each page.
- [Traces, logs, metrics, service map, API health, dashboard](https://powertraces.com/docs/observability_pages): What the core observability pages do.
- [Health, SLOs, incidents and alerts](https://powertraces.com/docs/health_incidents_alerts): Uptime checks, SLO error budgets, automatic incidents and alert rules.
- [Behavior (journeys, funnels, friction)](https://powertraces.com/docs/behavior): User-journey analytics from ordinary frontend spans.
- [Security](https://powertraces.com/docs/security): Threat detection from existing telemetry.
- [Reports and status pages](https://powertraces.com/docs/reports_status_pages): Scheduled reports and public status pages.
- [Organizations, workspaces, members and roles](https://powertraces.com/docs/teams_and_access): Who can see what, in the portal and in the app.
- [Billing, trials, quotas and retention](https://powertraces.com/docs/billing_and_limits): How plans, the daily ingest quota and retention work.
- [On-prem (self-hosted) edition](https://powertraces.com/docs/on_prem): Running PowerTraces on your own infrastructure.
- [Troubleshooting](https://powertraces.com/docs/troubleshooting): Fixes for the most common setup problems.

## Trace your AI calls and agents

PowerTraces reads the OpenTelemetry GenAI conventions, so any model call, tool call or agent run your libraries trace shows up on the AI page with its tokens, cost and steps. Connect your service first, then add the part for your provider.

- [Claude](https://powertraces.com/docs/ai/claude) (Anthropic): Set it up in the Claude Console with no code, add the Claude SDK's tracing in your app, or both.
- [OpenAI / ChatGPT](https://powertraces.com/docs/ai/openai) (OpenAI): Trace your app's calls to the OpenAI API, including the models behind ChatGPT, with the official OpenTelemetry instrumentation.
- [Gemini](https://powertraces.com/docs/ai/gemini) (Google): Trace Gemini calls made with the Google Gen AI SDK, on the Gemini API or Vertex AI.
- [.NET AI](https://powertraces.com/docs/ai/dotnet) (Microsoft.Extensions.AI): Any IChatClient (OpenAI, Azure OpenAI, Ollama and others) and Semantic Kernel or Agent Framework apps built on it.
- [Vercel AI SDK](https://powertraces.com/docs/ai/vercel) (Vercel): generateText, streamText and agents in the AI SDK, with any model provider it supports.
- [LangChain · LlamaIndex](https://powertraces.com/docs/ai/frameworks) (OpenInference): Agent frameworks traced with OpenInference; PowerTraces reads its attributes without any mapping.
- [Amazon Bedrock](https://powertraces.com/docs/ai/bedrock) (AWS): Bedrock Runtime calls (Converse and InvokeModel) made with boto3.
- [Other providers](https://powertraces.com/docs/ai/others) (Mistral · Ollama · Cohere · Groq): Any library that writes OpenTelemetry GenAI attributes works. OpenLLMetry covers many providers at once.