Future AGI docs

Add continuous LLM evaluation to your production AI system. Follow the quickstart and start evaluating in under ten minutes.

Future AGI in action

Future AGI instruments your existing LLM client calls. Set rubrics describing quality, and Future AGI scores each call against them in near-real time. The dashboard provides results, traces, and alerts.

The SDK keeps overhead low. Evaluation runs asynchronously from your response path, typically adding 2-4ms to the instrumented call. A separate worker pool handles scoring.

Explore quickly

Quickstart

Install the SDK, instrument a client call, set a rubric, and inspect dashboard traces. Begin in under ten minutes.

Integrations

Use LangChain, LlamaIndex, CrewAI, AutoGen, plus major LLM providers. One-line wrappers.

API reference

Parameter details for Python and TypeScript SDKs, REST API endpoints, and webhook payload schemas.

Rubric schema

Build evaluations with natural language, regex patterns, or both. See every rubric field and threshold type.

Supported providers

The Future AGI SDK supports clients from major LLM providers. Swap in the Future AGI wrapper, while leaving the rest of your code unchanged.

# Before
from openai import OpenAI
client = OpenAI()

# After (Future AGI wraps the client)
from futureagi import FutureAGI
client = FutureAGI(openai_api_key="sk-...", futureagi_api_key="fagi-...")

# All calls look identical to the original SDK
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Summarize this document."}],
    rubric={"criteria": "Factual, concise, under 3 sentences"}
)

Requirements

Python 3.9+ or Node.js 18+
Future AGI API key (Pro plan or higher; Dev plan includes limited trace retention)
Existing LLM provider API key