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 / C3-H3
What this task measures
The caller is on the line with Halverson & Reed for one locked request: Tight deadline but books intake. 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 a hard eligibility gates case because practice-area and jurisdiction gates have a required order that has to finish before advice or a case value. The agent succeeds only if it stays on that request, reaches the outcome by following the conflict screening then intake path, 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 $0.486
0:00 / 5:44 · audio · trace 803944 · click a span to play, drag to scrub, scroll to pan
Audio
Speech Audio
Agent Audio
Handoffs
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 did not match. Combined cannot pass unless every gate does.
Expected versus actual tool calls on this conversation.
Expected
Actual
Parameters
full_name: Nadine Carlisle
full_name: Nadine Carlisle
+ phone: 8055550167
Output
+ ok: true
Parameters
+ from: reception
+ to: screening
Output
None recorded.
None recorded.
Parameters
- opposing_party: Wesley Trombley
+ opposing_party: Unknown driver who pulled out of a driveway
Output
+ ok: true
Parameters
practice_area: auto_accident
practice_area: auto_accident
Output
+ ok: true
Parameters
practice_area: auto_accident
practice_area: auto_accident
state: CA
state: CA
Output
+ ok: true
Parameters
incident_date: 2024-09-15
incident_date: 2024-09-15
practice_area: auto_accident
practice_area: auto_accident
state: CA
state: CA
Output
+ ok: true
Parameters
+ from: screening
+ to: intake
Output
None recorded.
None recorded.
Parameters
incident_date: 2024-09-15
incident_date: 2024-09-15
practice_area: auto_accident
practice_area: auto_accident
state: CA
state: CA
- summary: Driveway pull-out collision.
+ summary: Caller reports an auto accident in California where another car pulled out of a driveway and struck the passenger side. Caller reports injuries from the crash and vehicle damage to the passenger side.
Output
+ ok: true
Parameters
channel: email
channel: email
Output
+ ok: true
Parameters
+ provider: Wesley Trombley
Output
+ ok: true
Parameters
+ provider: Wesley Trombley
Output
+ ok: true
Parameters
+ reason: Caller declined scheduling after intake and packet were sent; matter recorded and no further action requested.
Output
None recorded.
None recorded.
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Parameter miss: check_conflict, record_intake. Extra: request_records_authorization, request_records_authorization, end_call.
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
transfer_to_screening → transfer_to_intake
Actual path
transfer_to_screening → transfer_to_intake
The actual hangup matches the expected hangup.
Locked caller script. This is the technical definition. Copy it as written.
Open with exactly: "A car pulled out of a driveway and hit my passenger side." When asked who this would be against, say exactly: "It's Wesley Trombley." State is California. Date: September fifteenth, twenty twenty four. Stay until they confirm the intake is recorded and the new-client packet is on the file. A spoken promise without both on the file is not enough. Accept the packet by email if offered. If offered a free case evaluation or appointment, decline booking. Do not ask for a person. After they confirm intake is recorded and the packet is on the file, say exactly: "Thank you, that's all I needed." Do not book an evaluation. End the call.
