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-M4
What this task measures
The caller is on the line with Kestrel Electronics for one locked request: Fake 399 charge, no money sent. 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 taking the fraud desk 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 $0.245
0:00 / 2:53 · audio · trace 813162 · 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 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
+ to: fraud
Output
None recorded.
None recorded.
Parameters
Output
+ error_code: NO_SUCH_CHARGE
+ ok: false
Parameters
- email: [email protected]
+ email: [email protected]
phone: 541-555-0133
phone: 541-555-0133
Output
+ error_code: NO_OUTBOUND_CONTACT
+ ok: false
Parameters
amount: 399.99
amount: 399.99
channel: email
channel: email
claimed_brand: TechCrew
claimed_brand: TechCrew
- email: [email protected]
+ email: [email protected]
money_sent: false
money_sent: false
- payment_requested: none
+ payment_requested: asked to call a support number; no payment method stated
phone: 541-555-0133
phone: 541-555-0133
remote_access_given: false
remote_access_given: false
Output
+ ok: true
Parameters
+ reason: resolved scam contact report; no further help requested
Output
None recorded.
None recorded.
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Parameter miss: check_outbound_contact, report_scam_contact. 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_fraud
Actual path
transfer_to_fraud
The final-database gate failed, and no row-level difference was recorded between the expected hangup and the actual hangup.
Locked caller script. This is the technical definition. Copy it as written.
Open with exactly: "I got an email that my TechCrew subscription renewed for three hundred ninety-nine dollars and ninety-nine cents. Is that a real Kestrel charge?" Give your phone 541-555-0133 and email if asked. The amount in the email is $399.99. You have not sent money and have not given anyone device access. They did not ask you to buy store cards, send a gift card, or move money. Stay until they check the charge, check whether Kestrel contacted you, and file the report. After they say there is no such charge, that message did not come from Kestrel, nobody here emailed you, and they have given you next steps, thank them and hang up. Do not ask them to look up an order. Do not buy store cards or gift cards. 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.
