Post-call analytics everywhere
Every call on every tier gets opened: an AI summary and a QA score, on 100% of calls, not a sample. The capability is in the feature matrix today; what lands in August is the v2 API surface — an analysis object on the call and a webhook event when it is ready. Until then the summaries exist but are not queryable over v2.Warm transfer everywhere
Human handoff is part of every tier — a caller who asks for a person gets one. The August rollout wires the last mile:ended_reason: transferred starts emitting, and
transfer targets become configurable per agent. Ask your account contact to
enable it on your workspace as the rollout reaches you.
Model APIs
The ASR, TTS and LLM that power the call path, exposed as standalone endpoints — for teams that want the models without the telephony. Streaming ASR tuned for Indian telephone audio, the TTS voice catalogue, and the call-tuned LLM, each behind the samesk_live_ auth as the rest of v2.
Next-generation TTS
The next revision of the voice engine: lower first-audio latency and a step up in naturalness on Hindi and code-switched speech. It replaces the engine under every tier when it clears our quality bar — no API change, your calls just start sounding better.Orchestration platform
Node graphs get a canvas. Build, test and version call flows visually — the same graphst5 runs from JSON, authored by your ops team instead of your
engineers. Validation runs as you draw, and every published version is
immutable, so a live campaign never runs an unreviewed edit.
New voices
The catalogue grows monthly from August 2026. New voices appear in the voice gallery as they clear evaluation — listen tosampleAudio before you switch; voice choice moves answer-through rates more
than prompt wording does.