Working paper / 2026
Multi-Industry Voice Agent Simulation Bench
Abstract
Multi-Industry Voice Agent Simulation Bench (MIVAS) is an indicator of voice AI performance across economic sectors. MIVAS is comprised of tasks and verifiers with granular rewards executed in production-grade multi-agent RL environments across industries where voice AI is being adopted the fastest. Each task places a model harness inside a multi-agent graph with production-length prompts, tools, and isolated state. Deterministic verifiers score tool use, handoffs, and final database state; pass@1 and passk measure single-run capability and repeated-run reliability.
Clone the repository, pair a model harness with an industry task suite, and reproduce the evaluation environment.
Legal / C1-E1-SIG
What this task measures
The caller is on the line with Halverson & Reed for one locked request: Message for Daniel Okonkwo. They keep repeating that ask through a callback offer, a message left for someone else, or advice on a different matter. We included it as an easy reception and routing case because intake has to wait until the caller is classified; messages, wrong numbers, and represented parties are everyday reception work. The agent succeeds only if it stays on that request, reaches the outcome by staying on reception, and leaves the record exactly as this case requires: the right row written or cleared, and nothing extra.
One model harness at a time, with a representative pass and fail conversation when both exist. OpenAI Realtime 2.1 is the default view.
Recording
Conversation cost $0.131
0:00 / 1:49 · audio · trace 803801 · click a span to play, drag to scrub, scroll to pan
Audio
Speech Audio
Agent Audio
Tool Calls
Errors
Transcript
How a conversation passes
Combined pass is the logical and of three deterministic gates. Tools, handoff path, and final database state must all pass on the same conversation. One miss fails the call.
Combined
Fail
This sample fails because tools and final database state did not match. Combined cannot pass unless every gate does.
Expected versus actual tool calls on this conversation.
Expected
Actual
Parameters
+ full_name: Renee Whitmore
+ phone: 2145550163
Output
+ ok: true
Parameters
- for_whom: Daniel Okonkwo
+ for_whom: Daniel O’Connell
+ message: Caller: Renee Whitmore, callback number 214-555-0163. Message: The form I mailed arrived, and I have a question about page four.
Output
+ ok: true
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Parameter miss: take_message. Extra: lookup_caller.
Full industry architecture, same graph as the industry page. The left diagram marks the expected specialist path in green. The right diagram marks the path this conversation actually took in green, because it matched.
Expected path
stay on reception
Actual path
stay on reception
Rows that differ between the expected hangup and the actual hangup.
messages · 1
Expected
Actual
- for_whom: Daniel Okonkwo
+ for_whom: Daniel O’Connell
- message: Callback requested.
+ message: Caller: Renee Whitmore, callback number 214-555-0163. Message: The form I mailed arrived, and I have a question about page four.
Locked caller script. This is the technical definition. Copy it as written.
Open with exactly: "I need to leave a message for Daniel Okonkwo." You are calling about forms Daniel mailed you. You do not want a new matter or an appointment. When asked what the message is, say exactly: "Tell him the forms he mailed arrived and I have a question about page four." If asked about a new matter, say exactly: "No, just the message please." Once finished, say exactly: "Thank you, that's all I needed." and end the call.
