# Blog

> Research notes, product updates and engineering write-ups from the team building Tazo Labs.

URL: https://tazolabs.com/blog

## Posts

- [Introducing Tazo: the audio decision model for voice agents](https://tazolabs.com/blog/introducing-tazo-audio-decision-model): Tazo is an audio decision model for live voice agents. Raw call audio in, a calibrated probability for each option out, in under 200 ms. API is live. (2026-11-11)
- [How audio decision models make graph agents reliable](https://tazolabs.com/blog/reliable-graph-agents-voice-ai): Graph agents fail when nobody can say which state a call was in or why it moved. How an audio decision model makes every transition scored and logged. (2026-10-18)
- [Stop making your personal AI agent transcribe first](https://tazolabs.com/blog/personal-ai-agent-voice-commands): Build a voice-controlled AI agent without a transcript. An audio decision model picks the action, and your thresholds decide to act, confirm or ignore. (2026-10-13)
- [Post-call analysis: query your calls, not just transcripts](https://tazolabs.com/blog/post-call-analysis-voice-ai): Most post-call analysis reads transcripts and misses tone, timing and what was actually said. How an audio decision model lets you query every call recording. (2026-10-13)
- [What the heck is an audio decision model?](https://tazolabs.com/blog/what-is-an-audio-decision-model): An audio decision model listens to audio and answers a question about it, with a probability. What it does, and over 30 use cases in voice AI. (2026-10-04)
- [Tazo vs Jev: why voice AI needs an audio decision model](https://tazolabs.com/blog/tazo-vs-jev-audio-decision-model): An audio decision model hears a call and picks the next action. How Tazo compares with Jev and AudioJev, and why voice AI agents need one. (2026-10-03)
- [Ask what the moment needs](https://tazolabs.com/blog/ask-what-the-moment-needs): Voice AI agents decide every few hundred ms: respond, wait, yield or escalate. Why most get it wrong, and what an audio-native decision model changes. (2026-10-02)
- [Speculative tool calling: the next big thing in voice AI](https://tazolabs.com/blog/speculative-tool-calling-voice-ai): Tool calls are where voice agent latency goes. Speculative tool calling starts safe lookups while the caller is still talking, without changing behavior. (2026-10-02)
- [The missing piece in the voice AI stack: decision models](https://tazolabs.com/blog/voice-ai-decision-models): Voice AI stacks have great models for STT, LLM and TTS. The layer that decides when to speak is still a silence timer. Why it needs a decision model. (2026-10-02)
