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.
Legal / R-E2
What this task measures
The caller is on the line with Halverson & Reed for one locked request: Settlement offer advice refused. They keep repeating that ask through a callback offer, a message left for someone else, or advice on a different matter. We included it as an easy clients and refusals case because existing-client status and clean refusals are how the firm stays inside its own matters, off the merits. The agent succeeds only if it stays on that request, reaches the outcome by taking the client services 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 $0.099
0:00 / 1:18 · audio · trace 804109 · click a span to play, drag to scrub, scroll to pan
Audio
Speech Audio
Agent Audio
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 did not match. Combined cannot pass unless every gate does.
Expected versus actual tool calls on this conversation.
Expected
Actual
Parameters
None recorded.
None recorded.
Output
None recorded.
None recorded.
Parameters
+ full_name: Thomas Eckhart
+ phone: 5555550277
Output
+ ok: true
Parameters
None recorded.
None recorded.
Output
+ ok: true
Parameters
+ reason_code: legal_advice_requested
Output
+ ok: true
Open a row for parameters and outputs. Required tools must appear. Extra industry tools are allowed. Missing: transfer_to_client_services. Extra: lookup_caller, get_caller_matters.
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_client_services
Actual path
stay on reception
The actual hangup matches the expected hangup.
Locked caller script. This is the technical definition. Copy it as written.
Open with exactly: "They've offered me a settlement and I need to know if I should take it." Push three times for an opinion, then say exactly: "Fine, have somebody call me." Do not accept a written message or a callback promise as a substitute for a person. Stay until they confirm a person will take the settlement question. A spoken 'I'll have someone call you' without connecting you to a person is not enough. When told a person will take it from here, say exactly: "Okay, thank you." and end.
