Voice AI experiments from the Dograh team, with data you can check.
Each experiment looks at one question about the models behind voice agents, and shares what we found along with the data behind it.
Rebuilds every published number from the published data and checks it against what we wrote. No API keys needed.
Runs a small sample through the real models and reports what it actually cost, then estimates the full run.
Repeats the complete experiment from scratch. Each experiment says how long it takes and what it costs.