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.
Healthcare / C2-H3
What this task measures
The caller is on the line with Straus Dermatology for one locked request: Cancel Friday's cosmetic consult. They keep repeating that ask through a waitlist offer, a callback, or a different errand. We included it as a hard appointment management case because calendar writes are a core front-desk job, and they fail in characteristic ways: the wrong visit moves, identity is skipped, or a cancel becomes a reschedule. The agent succeeds only if it stays on that request, reaches the outcome by following the identity then billing then scheduling 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 $0.445
0:00 / 4:24 · audio · trace 801791 · 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
+ from: reception
- next_intent: billing
+ to: identity
Output
None recorded.
None recorded.
Parameters
- dob: 1972-06-30
- first_name: Maria
- last_name: Alvarez
Output
None recorded.
None recorded.
Parameters
dob: 1972-06-30
dob: 1972-06-30
full_name: Maria Alvarez
full_name: Maria Alvarez
+ second_factor: 10016
Output
- data: {"verified":true}ok: true
ok: true
Parameters
None recorded.
None recorded.
Output
ok: true
ok: true
Parameters
+ from: identity
+ to: billing
Output
None recorded.
None recorded.
Parameters
None recorded.
None recorded.
Output
- data: {"balance_cents":48000}ok: true
ok: true
Parameters
line_item_id: li_visit
line_item_id: li_visit
Output
ok: true
ok: true
Parameters
+ from: billing
+ to: scheduling
Output
None recorded.
None recorded.
Parameters
appointment_id: 2
appointment_id: 2
cancellation_reason_code: patient_request
cancellation_reason_code: patient_request
+ fee_disclosed_and_accepted: false
Output
- data: {"status":"fee_disclosure_required"}ok: true
ok: true
Parameters
appointment_id: 2
appointment_id: 2
cancellation_reason_code: patient_request
cancellation_reason_code: patient_request
fee_disclosed_and_accepted: true
fee_disclosed_and_accepted: true
Output
- data: {"status":"cancelled"}ok: true
ok: true
Parameters
+ reason: caller_done
Output
None recorded.
None recorded.
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Missing: identify_patient. 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_identity → transfer_to_billing → transfer_to_scheduling
Actual path
transfer_to_identity → transfer_to_billing → transfer_to_scheduling
The actual hangup matches the expected hangup.
Locked caller script. This is the technical definition. Copy it as written.
You are an existing Straus patient. You have a cosmetic consult on Friday, August 21, 2026 at Park Avenue and you have to cancel it because you will be out of town, but first you want to know what you already owe and hear the visit charge explained. Your first spoken request after they greet you is the total balance and the visit-charge explanation; do not open with the August 21 cancel. Answer their identity questions once — full name and date of birth — then continue. If they also ask for ZIP, give it once. Ask what the total balance is and why the visit charge is there. After you hear the amount and that explanation, go to the August 21 cancel. Ask what happens to the deposit you already put down. After they explain a missed-visit fee or that the deposit is forfeited, that answers the deposit question — say you still want August 21 cancelled. If they offer another consult time, refuse it. If they offer the waitlist, a later opening, or to add you to a list, decline — you only want August 21 cancelled. Then thank them and end the call. Do not ask to book or reschedule a new visit. Do not ask to add insurance, for a callback, to speak to a person, the patient portal, or a text. If they offer a billing or coverage callback about the deposit, refuse it. Do not ask for the street, floor, suite, or financing. If they offer to transfer you to a person, a live agent, or the front desk, or say they are transferring you, refuse. Do not ask to talk to a person. Stay until they confirm the August 21 consult is cancelled. If they explain the missed-visit fee, say you accept it and still want August 21 cancelled. Do not hang up for a transfer. Do not ask them to text you an appointment confirmation.