Frequently Asked Questions

Learn how Spanora helps teams understand AI behavior, control model spend, and debug agent runs with OTEL-native tracing.

Why teams choose Spanora

  • Understand exactly what happened in every AI execution with full trace timelines.
  • Pinpoint expensive prompts, models, and tool calls with span-level token and cost data.
  • Debug faster by inspecting raw prompt inputs, outputs, and failures in one place.
  • Stay flexible with OTEL-native ingestion that works with your existing telemetry stack.

Need implementation details? Visit the docs.

About Spanora

What is Spanora?
Spanora is an OTEL-native observability platform for AI executions. It helps you see LLM calls, tool calls, prompts, outputs, token usage, cost, and outcomes so you can answer what happened, why it happened, and how much it cost.
Who is Spanora for?
Spanora is built for teams shipping AI products and agentic workflows. If you need to troubleshoot failures, understand behavior, and control model spend, Spanora gives your engineers and product teams the context they need.
What can I answer with Spanora?
You can answer the core execution questions quickly: what happened, which agent ran, which prompt or tool caused a failure, why cost increased, and which users, sessions, or organizations drive usage.

Benefits and outcomes

How does Spanora help reduce debugging time?
Instead of searching across logs and multiple dashboards, Spanora gives you one timeline per execution with span-level details. You can move directly from a failed trace to the exact prompt, model response, or tool error.
How does Spanora help with AI cost control?
Spanora captures tokens and cost per span, then rolls them up to trace-level views. You can identify expensive prompts, regressions after releases, and high-cost users or organizations before spend surprises you.
What makes Spanora different from vendor-specific tools?
Spanora is built on OpenTelemetry, not a closed instrumentation format. That keeps your data portable and your stack flexible, even as your models and frameworks evolve.

Setup and integrations

How do I get started?
Create an account, generate an API key, and point your OTEL exporter to Spanora's OTLP HTTP endpoint. You can start with existing telemetry and improve signal quality over time.
Do I need the Spanora SDK?
No. Spanora works with raw OTEL data and does not require a proprietary SDK. The SDK is optional and provides developer ergonomics like consistent attribute helpers.
Which frameworks and providers can I use?
Any framework that emits OpenTelemetry traces can work with Spanora. Common setups include OpenAI, Anthropic, Vercel AI SDK, and LangChain, but you are not limited to a specific language, runtime, or model provider.
What can I inspect in a trace?
You can inspect timeline order, latency, status, model and provider, prompts, outputs, tool execution details, tokens, and cost. Spanora keeps span detail front and center so root causes are visible quickly.

Pricing and plans

Is there a free plan?
Yes. You can start on a free plan with 1,000 and 24 hours retention, with no credit card required.
What are the paid options?
Paid plans include Starter (50,000 with 30-day retention) and Pro (500,000 with 90-day retention), with Enterprise available for larger workloads.
What counts as a span?
A span is a single unit of work inside a trace, such as one LLM call, tool call, or operation step. Usage limits are based on the number of spans ingested.
How does retention work?
Retention controls how long trace data stays queryable. Current plan retention windows are 24 hours on Free, 30 days on Starter, and 90 days on Pro.

Data and privacy

What data does Spanora store?
Spanora stores the telemetry you send via OTLP, including spans and related metadata such as prompts, outputs, tool details, token usage, and cost when provided. You control what is sent from your instrumentation.
Does Spanora automatically redact PII?
No. Spanora does not automatically redact sensitive content. You should filter or redact confidential data in your app or telemetry pipeline before exporting traces.
Can I export my data?
Yes. Your telemetry data remains yours. If you need an export workflow, contact support and we can help with a path that fits your plan and volume.