Best viewed on desktop

The Playground is a multi-pane code environment designed for larger screens. Open it on your laptop or desktop for the full experience.

pip install electripy-ai View the quickstart instead →

ElectriPy AI · Playground

Pick a pattern. Read the code. See it run — no setup required.

pip install electripy-ai

Running in production?ElectriPy Cloud →

Basic AI Runtime

beginner

Calling an LLM directly means zero visibility, zero retry logic, and a single point of failure. The ElectriPy runtime wraps every call in a fully instrumented, policy-aware execution context.

Docs

L04 Model Runtime

LLM GatewayRuntime TracingToken Counting

code

$ pip install electripy-ai

from electripy import ai response = await ai.complete( model="gpt-4o-mini", messages=[{"role": "user", "content": "Explain AI observability in 30 words."}], ) print(response.content) print(f"Tokens used: {response.usage.total_tokens}")

output

✦ Simulated outputNo API key or network required

output

{
  "content": "AI observability means tracking every LLM call—inputs, outputs, latency, cost, and errors—so you can debug, optimise, and govern AI reliably in production.",
  "usage.prompt_tokens": 24,
  "usage.completion_tokens": 31,
  "usage.total_tokens": 55,
  "model": "gpt-4o-mini",
  "latency_ms": 312
}

runtime events

1
Request received by runtime
2
Policy pre-flight checkALLOW
3
LLM call dispatched
4
Response received31 tokens
5
Span closed — latency 312 ms

Key takeaway: Every call goes through policy, tracing, and cost tracking automatically — no extra code required.

Why it matters: Your team gets full visibility into every AI decision — what was asked, what came back, how long it took, and what it cost. No more black boxes, no more surprise bills.