Test guides by platform
How to test a voice agent built on a named platform. Each guide covers which defaults to check and how to confirm them on a real call.
Short answer. A platform guide tells you which of its defaults to check on a real call before you trust them. Turns, transfers, recording and disclosure are the places to look first.
Check the live call. Settings pages describe intent. The call shows what the platform really did.
What applies on any platform
The two builds behind most platforms and what each means for testing.
Checks that behave the same whichever platform runs the call.
Guides by platform
Each guide is written from the platform's own public documentation and links the page behind every setting it names.
Evals, simulations, call logs and webhooks, then the speech and transfer settings to check on a real call.
Simulation, web and phone tests, call history and post-call analysis.
Simulation tests, evaluation criteria, data collection and post-call webhooks.
Unit tests with a judge, simulations, Agent Console and real calls over SIP.
Evals, metrics, tracing and real calls over a phone transport.
Choosing a platform
Retell, Bland, Synthflow, ElevenLabs Agents, LiveKit and Pipecat, and what to test before you switch.
Build, control, telephony, latency claims, pricing and testing hooks, side by side.
How Vapi, Retell and Bland bill a call, with a worked 3-minute example.
Whatever the platform, a typed test and a real call find different faults. See text vs voice testing and call traces and OpenTelemetry.