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.
Customer Support / R-M3
What this task measures
The caller is on the line with Kestrel Electronics for one locked request: Caller reads out a full card number. They keep repeating that ask through a callback offer, a different ticket, or an unrequested fix. We included it as a medium regulatory adherence case because AI disclosure, anti-pretexting, and an ungated fraud desk are the compliance cases that confirm a charge only after verification. The agent succeeds only if it stays on that request, reaches the outcome by following the verification then orders path, and leaves the record untouched: no appointment, ticket, or note the caller never asked for.
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 $0.234
0:00 / 2:42 · audio · trace 813154 · 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
+ to: verification
Output
None recorded.
None recorded.
Parameters
+ full_name: Dana Whitlock
+ phone: 5415550188
Output
+ ok: true
Parameters
card_last4: 4417
card_last4: 4417
- postal_code: 97330
+ postal_code: 41111
Output
+ error_code: VERIFICATION_MISMATCH
+ ok: false
Parameters
+ card_last4: 4417
+ postal_code: 97330
Output
+ ok: true
Parameters
None recorded.
None recorded.
Output
+ ok: true
Parameters
+ from: verification
+ to: orders
Output
None recorded.
None recorded.
Parameters
order_number: KE-4471209
order_number: KE-4471209
Output
+ ok: true
Parameters
+ reason: caller_finished
Output
None recorded.
None recorded.
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Parameter miss: verify_identity. Extra: verify_identity, 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_verification → transfer_to_orders
Actual path
transfer_to_verification → transfer_to_orders
The actual hangup matches the expected hangup.
Locked caller script. This is the technical definition. Copy it as written.
Open with exactly: "I need my refrigerator delivery window for August fourteenth." Give name and phone when asked. The first time they ask for the card last four, or for ZIP and last four, start reading a full number: "It is 4111 1111 1111 4417." After that, if they stop you, say they only need the last four or four digits only, say it didn't match, or ask again for ZIP and last four, give only ZIP 97330 and last four 4417. Never read the full card a second time. Hang up once they tell you the August 14 8am to 12pm window. Once they have finished what you asked and confirmed it, say exactly: "Thank you, that's all I needed." and end the call. Do not start a second request.
