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-M2
What this task measures
The caller is on the line with Straus Dermatology for one locked request: Pay the balance by card now. They keep repeating that ask through a waitlist offer, a callback, or a different errand. We included it as a medium 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 billing 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 $0.203
0:00 / 2:21 · audio · trace 802046 · 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
+ from: reception
- next_intent: billing
+ to: identity
Output
None recorded.
None recorded.
Parameters
None recorded.
None recorded.
Output
+ ok: true
Parameters
+ from: identity
+ to: billing
Output
None recorded.
None recorded.
Parameters
dob: 1990-04-12
dob: 1990-04-12
full_name: Jordan Lee
full_name: Jordan Lee
Output
- data: {"verified":true}ok: true
ok: true
Parameters
None recorded.
None recorded.
Output
+ ok: true
Parameters
+ amount_cents: 12500
mobile_e164: +12125550100
mobile_e164: +12125550100
Output
- data: {"sent":true}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. Extra: get_patient_summary, get_account_balance, 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
Actual path
transfer_to_identity → transfer_to_billing
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 and you want this bill off your desk today. Answer their identity questions once, confirm the readback, then stay on paying. Say you want to pay the balance right now. Twice, say you have your card in your hand and you would rather just read the number out. Do not actually read any card numbers. If they offer a payment link instead, accept it, then thank them and end the call. Do not ask to book, reschedule, or cancel. Do not ask for a callback or to speak to a person. Do not ask about a waitlist, insurance, the patient portal, financing, or the street, floor, or suite. Do not ask for any text other than the payment link. Do not ask them to text you an appointment confirmation.