policybazaar.ai assistant
policybazaar.ai team · Chat · In-house
Live
Conversations
521
from files
Scoring new traffic
Every 15 min
last ran just now
Uploaded · 521 conversations from files
Upload another file
Up to 5,000 conversations per file. Not masked on the way in (the SDK masks in the app); Dvar masks names and numbers before any model reads them.
One conversation per line: {"id": "c1", "messages": [{"role": "user", "content": "…"}, {"role": "assistant", "content": "…"}]}
Go live
Connect the agent and every new conversation is scored automatically, every 15 minutes. No more exports.
No key yet. One key per workflow, for the gateway and the SDK.
- 1
Point the agent's OpenAI-compatible client at Dvar
from openai import OpenAI client = OpenAI(base_url="http://localhost:4000/v1", api_key="<your Dvar key>")
- 2
Send each conversation's id, so its turns group into one conversation
client.chat.completions.create(model="dvar-fast", messages=messages, extra_body={"metadata": {"session_id": conversation_id}}) - 3
Conversations appear here within about 15 minutes.