LLM Observability Platform · Built for Teams

The LLM observability
platform for your team.

Shared dashboards, per-tenant cost tracking, team alerts, and compliance packs — one platform for every LLM call across all your projects. MIT SDK. No proxy. Free to start.

OpenAI Anthropic Gemini Bedrock LangChain CrewAI LlamaIndex

What a platform gives you

Beyond the SDK — everything your team needs.

The MIT SDK handles instrumentation. The platform handles everything else: shared visibility, team workflows, cost accountability, and compliance — across every project and every developer.

Shared trace dashboards

Every developer on your team sees the same traces. No more 'works on my machine' — any LLM call, any agent step, searchable and shared by default.

Per-tenant cost tracking

Break LLM cost down by customer, by project, by operation. Know which tenant is driving your bill before the invoice arrives.

Compliance across projects

Toggle HIPAA, FDCPA, FINRA, and GDPR compliance packs per project from the dashboard. Legal changes a rule; the SDK picks it up on next startup.

Team alerts

Get notified on cost spikes, hallucination regressions, and latency anomalies. Route alerts to Slack or PagerDuty so the right person is paged.

Multi-project workspace

One workspace, many projects. Switch between staging, production, and customer environments without changing API keys or losing context.

Audit-ready access control

Role-based access so your compliance officer has read-only access and your engineers have full trace access. SAML/SSO on Enterprise.

Two lines per service.

One platform. Every service your team ships.

Add peekr.instrument() to the top of each service. The platform groups traces by project and tenant automatically — no extra configuration beyond the project name.

  • One workspace for every project and environment
  • Per-tenant breakdown built in — pass tenant_id and it just works
  • Compliance packs controlled from the dashboard — no code change
  • Zero-latency overhead — spans export on a background thread
service.py
import peekr

# Same two lines — every service, every project
peekr.instrument(
  tenant_id=request.user_id,  # per-customer cost
  exporter=peekr.HTTPExporter(
    endpoint="https://peekr.starkspherelabs.com",
    api_key="pk_live_…",
    project="search-service-prod",
  ),
  compliance=["HIPAA"],  # set in dashboard
)

# Platform groups traces by project + tenant.
# Your team sees everything in one workspace.

Platform vs. DIY observability

What you get on day one vs. what you'd build in six months.

CapabilityPeekr PlatformDIY / Custom logging
Setup 2 lines of PythonWeeks of instrumentation work
Cost tracking Per-tenant, per-operation, built-inCustom aggregation pipeline
Hallucination detection Sentence-level, includedRequires separate eval harness
Compliance enforcement Toggle from dashboard, 17 packsManual regex rules, no audit log
Team dashboards Shared by defaultBuild your own Grafana/Kibana
Alerts Cost, latency, hallucination thresholdsCustom alerting pipeline
Maintenance Peekr maintains the platformYou maintain everything

Start observing in two lines.

Free up to 10k spans per month. No credit card. Works with every LLM SDK your team already uses — OpenAI, Anthropic, Gemini, Bedrock, LangChain.