/helicone-observability

Route LLM API calls through Helicone for cost tracking, latency monitoring, and per-user attribution across all products.

Architecture

App → Helicone Gateway (proxy) → LLM Provider (Anthropic, OpenAI, etc.)
                ↓
    Helicone Dashboard (cost, latency, properties)

Helicone is a transparent proxy. One URL + header change per provider. No SDK lock-in.

Multi-Product Strategy

Helicone has ONE dashboard per account. Segmentation is via custom properties, not separate projects.

Mandatory properties for every LLM call: