Skip to content

Coalex AI Documentation

AI governance for enterprises in regulated markets. Track, Protect, and Trust your AI agents.

Coalex solves "AI Pilot Purgatory" -- enterprises wasting billions on AI pilots that never reach production. Our platform provides AI compliance and accuracy through democratizing AI engineering best practices: observe your AI agents, understand their behavior, and apply real-time guardrails.

  • Track

    Are your AI agents delivering value?

    Auto-instrument 14+ LLM frameworks. See every token, latency, and cost across your entire AI fleet -- with zero code changes.

    Learn more

  • Protect

    Can you trust your AI agents?

    Evaluate outputs against risk policies, escalate dangerous responses, and resolve with human-in-the-loop governance.

    Learn more

  • Trust

    Can others trust your AI agents?

    EU AI Act, ISO 42001, and SOC 2 compliance out of the box. Public trust center for transparency with customers and regulators.

    Learn more

Get started in 5 lines of code

import coalex
from openai import OpenAI

# 1. Connect to Coalex
coalex.register(
    endpoint="https://your-org.coalex.ai",
    api_key="your-api-key",
    service_name="my-agent",
)

# 2. Auto-instrument all supported frameworks
coalex.auto_instrument()

# 3. Wrap your agent invocation with context
with coalex.coalex_context(agent_id="support-bot", request_id="req-123"):
    client = OpenAI()
    response = client.chat.completions.create(
        model="gpt-4o",
        messages=[{"role": "user", "content": "Summarize this claim."}],
    )
    print(response.choices[0].message.content)

Every LLM call, token count, and latency metric is automatically captured and sent to your Coalex dashboard -- no manual instrumentation required.

Next steps

How it works

graph LR
    A[Your AI Agent] -->|OTLP traces| B[Coalex Collector]
    B --> C[Transformer]
    C -->|Enrichment| D[DuckLake]
    A -->|evaluate| E[Risk Engine]
    E -->|escalate| F[Human Review]
    F -->|resolve| A
    D --> G[Dashboard & Trust Center]
  1. Instrument your agent with the Coalex SDK -- automatic or manual
  2. Collect traces via the OTLP-compatible collector
  3. Enrich with derived metrics: PII detection, cost estimation, CO2 impact
  4. Evaluate outputs against your risk policies
  5. Escalate risky outputs for human-in-the-loop review
  6. Report compliance posture through the public trust center