We Tested 7 AI Assistants on Real Phone Calls — Most Fail at Pharmacies and Voicemails
Blocker03 · reddit · 2026-11-07
A team spent two weeks running the same 5 errands (restaurant booking, store hours, medical appointment, pharmacy refill, voicemail callback) through 7 AI assistants via Bland's outbound calling infra, scoring completion, latency, hangups, and transcript quality.
Key findings:
- ChatGPT handled booking and hours but fumbled the pharmacy call and got confused on hold
- Claude went furthest on appointment confirmation but failed the voicemail callback entirely
- Gemini was inconsistent: two tasks passed, three went silent
- Muse passed three tasks but had the worst latency
- Poke aced the voicemail callback yet failed booking outright
- Biggest surprise: businesses hang up on any pause over 3 seconds
More from coding & agent
- Dev's agent runs full Stripe integration rehearsals, tokens rival the app itself — tristanbob · 2026-11-07
- Alibaba's Qoder launches stealth model Qoduck, free for individual users — hey_abusiddik · 2026-11-07
- Grok Bot Slack integration spotted, coming in macOS 1.1.0 build — testingcatalog · 2026-11-07
- LLM4Code 2027 at ICSE Dublin opens CFP with Nov 13 deadline — LingmingZhang · 2026-11-07
- David Khourshid: Telling agents what you want beats telling them what to do — DavidKPiano · 2026-11-07
- Google's own coding agent got a dev's Cloud project banned by spamming Google's API — RexDouglass · 2026-11-07