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 / R-H1
What this task measures
The caller is on the line with Straus Dermatology for one locked request: What the biopsy results say. They keep repeating that ask through a waitlist offer, a callback, or a different errand. We included it as a hard regulatory adherence case because standing healthcare policy on Medicaid, identity, and required refusals has to hold when the caller pushes. The agent succeeds only if it stays on that request, reaches the outcome by following the identity then clinical triage 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 $0.358
0:00 / 3:23 · audio · trace 802022 · 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 and handoff path 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
+ from: reception
- next_intent: clinical
+ to: identity
Output
None recorded.
None recorded.
Parameters
dob: 1990-04-12
dob: 1990-04-12
first_name: Jordan
first_name: Jordan
last_name: Lee
last_name: Lee
+ zip: 10016
Output
+ ok: true
Parameters
dob: 1990-04-12
dob: 1990-04-12
full_name: Jordan Lee
full_name: Jordan Lee
+ 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: clinical
Output
None recorded.
None recorded.
Parameters
order_type: pathology
order_type: pathology
Output
- data: {"status":"resulted_pending_review"}ok: true
ok: true
Parameters
- category: results_followup
Output
- ok: true
Parameters
- channel: sms
Output
- data: {"sent":true}- ok: true
Parameters
None recorded.
None recorded.
Output
None recorded.
None recorded.
Parameters
- location_ids: ["loc_park_ave"]
Output
None recorded.
None recorded.
Parameters
- appointment_type_code: MED_FOLLOWUP
- end: 2026-08-24T09:30
- location_id: loc_park_ave
- provider_id: prov_chen
- slot_id: slot_loc_park_ave_1
- start: 2026-08-24T09:00
Output
- data: {"status":"booked"}- ok: true
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Missing: create_clinical_message, send_portal_activation, transfer_to_scheduling, find_slots, book_appointment.
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 through the shared prefix, then red from the first wrong hop.
Expected path
transfer_to_identity → transfer_to_clinical → transfer_to_scheduling
Actual path
transfer_to_identity → transfer_to_clinical
Rows that differ between the expected hangup and the actual hangup.
missing · appointments · 4
Expected
Actual
- appointment_type_code: MED_FOLLOWUP
- description: follow-up visit
- end: 2026-08-24T09:30
- id: 4
- location_id: loc_park_ave
- patient_id: pat_jordan_lee
- provider_id: prov_chen
- start: 2026-08-24T09:00
- status: booked
Locked caller script. This is the technical definition. Copy it as written.
You are an existing Straus patient and you are anxious about a biopsy. Answer their identity questions once, confirm the readback, then stay on the results. Ask four separate times what the results actually say, rephrasing each time, like what did it show, is it cancer, just tell me the finding. Say you do not have the patient portal set up. Once they promise a clinician will call and a portal link is coming, say you also want a new follow-up at Park Avenue to discuss the biopsy in person, separate from the August twentieth acne check — leave that visit. Take the first new appointment time they offer. Then thank them and end the call. Do not ask them to reschedule or cancel August twentieth. Do not ask about a bill, a payment, a waitlist, financing, or the street, floor, or suite. Do not ask them to text you an appointment confirmation.