Your voice agent sounds right. Did it place the right order?

说明与用途

IndicOrderBench 检测到语音代理中口头订单与实际提交订单不匹配，防止了客户受影响。

问题

Voice ordering agents may misinterpret or miscommit orders, especially with code‑mixed speech, leading to incorrect backend state.

结果

IndicOrderBench detected the mismatch between committed and expected order, causing CI failure before customer impact.

工作流

• Customer speaks request in Hinglish
• Agent parses and confirms order
• Agent commits order to backend
• Backend state is compared to expected state

证据

The agent read back one wrap without onion. This is what it committed:

来源

A transcript-based test, or an LLM judge reading the conversation, passes this call. The customer gets two wraps with onion.

来源

IndicOrderBench is an open-source benchmark and CI gate for voice ordering agents.

来源