Route agent frameworks through Jents

LangChain, CrewAI, LlamaIndex, AutoGen and friends all let you set a custom base URL on the model. Set it once and the whole agent — every tool call, sub-step, and retry — routes through Jents.

1. Create the source in Jents

Capture AI source → "AI you build or run" → "Yes — I can point it at a URL." Name it, create it, then Provision the key on the Keys tab. Copy your gateway URL + key.

2. Set the base URL on the model

LangChain (Python):

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="gpt-4o",
    base_url="https://YOUR-GATEWAY-URL",
    api_key="YOUR-AGENT-KEY",
)

CrewAI · LlamaIndex · AutoGen: each wraps an OpenAI-compatible client — set the same base_url and api_key on the LLM you pass in (e.g. OpenAI(base_url=…, api_key=…) for LlamaIndex; the llm config for CrewAI and AutoGen).

Set it on the LLM once and every agent, tool, and retry built on it routes through Jents — no per-call changes.

3. Run it and confirm

Run the agent once → the source's Keys tab shows "Working ✓" and spend lands in Observability, with multi-step runs grouped together.