How we hit sub-second pickup without sacrificing voice quality
A walkthrough of the latency budget that gets a real person on the line in under 900ms — TTFB, model warm pools, and the tricks we play with backchannel sounds.
Voice latency, agent reliability, telephony pitfalls, and the operator stories behind the product. No SEO filler — just the stuff we'd want to read.
A walkthrough of the latency budget that gets a real person on the line in under 900ms — TTFB, model warm pools, and the tricks we play with backchannel sounds.
Why we ship both, when each one wins, and how we keep the caller from hearing dead air during the handoff.
Three rebill structures we've watched succeed, the margins to expect, and how to position the upsell at month two.
Most hallucinations come from the knowledge base, not the LLM. A simple framework for cleaning yours up in an afternoon.
What we learned shipping Nova, Iris, Atlas, Marco, and Juno — and why pacing matters more than timbre.
We pulled six months of customer data to find out which industries leave the most money on the floor between 6pm and 8am.