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 / C4-H2
What this task measures
The caller is on the line with Halverson & Reed for one locked request: Consumer intake, declines booking. 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 intake and documents case because taking a narrative and a document packet is representative plaintiff work that starts after the conflict-before-facts rule has held. 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 $0.337
0:00 / 4:03 · audio · trace 803996 · 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
Pass
All three gates matched, so the conversation counts as a pass.
Expected versus actual tool calls on this conversation.
Expected
Actual
Parameters
full_name: Estelle Kowalczyk
full_name: Estelle Kowalczyk
+ phone: 9165550128
Output
+ ok: true
Parameters
+ from: reception
+ to: screening
Output
None recorded.
None recorded.
Parameters
opposing_party: Midtown Appliance
opposing_party: Midtown Appliance
Output
+ ok: true
Parameters
practice_area: consumer
practice_area: consumer
Output
+ ok: true
Parameters
practice_area: consumer
practice_area: consumer
state: CA
state: CA
Output
+ ok: true
Parameters
incident_date: 2026-01-05
incident_date: 2026-01-05
practice_area: consumer
practice_area: consumer
state: CA
state: CA
Output
+ ok: true
Parameters
+ from: screening
+ to: intake
Output
None recorded.
None recorded.
Parameters
incident_date: 2026-01-05
incident_date: 2026-01-05
practice_area: consumer
practice_area: consumer
state: CA
state: CA
- summary: Defective space heater.
+ summary: Caller reports purchasing a defective space heater from Midtown Appliance in California. The heater overheated and caused damage. Caller wants to file a claim.
Output
+ ok: true
Parameters
channel: email
channel: email
Output
+ ok: true
Parameters
+ reason: Caller prefers to review the intake paperwork before booking a free case evaluation. Packet has been sent by email. Intake recorded.
Output
None recorded.
None recorded.
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Parameter miss: record_intake. Extra: 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 store sold me a defective space heater." When asked who this would be against, say exactly: "It's Midtown Appliance." California. January fifth, twenty twenty six. If offered the packet, say exactly: "Email is fine." If offered an appointment, say exactly: "Not yet, let me read the paperwork first." Stay until they confirm the intake is on the file and the packet has been emailed. A spoken promise to send it without those writes is not enough. Once both are confirmed on the file, say exactly: "Thank you, that's all I needed." and end the call.
