https://thiyaga.xyz/categories/voice-agents/ Recent content in Voice Agents on Blog by Thiyagaraj T Hugo en Wed, 28 Oct 2026 00:30:00 +0530 https://thiyaga.xyz/posts/conversation-quality-metrics/ Wed, 28 Oct 2026 00:30:00 +0530 https://thiyaga.xyz/posts/conversation-quality-metrics/ <p><em>Say you run an agent that does 20-minute phone screens. It greets the candidate, asks about their last two roles, digs into one project, answers questions about the job, and wraps up. Thousands of calls a day. You change one line in the prompt. Did the calls get better or worse?</em></p> <p><em>This post is how we answer that question.</em></p> <hr> <h2 id="1-what-does-not-work">1. What does not work</h2> <p><img src="https://thiyaga.xyz/images/cq-no-golden.svg" alt="A task eval diffs one output against one reference; a conversation has three good next questions and nothing to diff against"></p> https://thiyaga.xyz/posts/agentic-voice-pipelines/ Mon, 28 Sep 2026 11:30:00 +0530 https://thiyaga.xyz/posts/agentic-voice-pipelines/ <p>The fastest way to give a product a voice today is a <strong>speech-to-speech model</strong>: one realtime model that hears audio and speaks audio directly, like OpenAI&rsquo;s Realtime API or Gemini Live. And for simple voice interaction it is genuinely good. Ask your app a question, dictate a note, chat hands-free while driving. One API call, natural prosody, low latency. When voice is a supplementary experience layered on top of a product, this is the right tool, and you can stop reading here.</p> https://thiyaga.xyz/posts/speculative-routing/ Mon, 03 Aug 2026 00:30:00 +0530 https://thiyaga.xyz/posts/speculative-routing/ <p><em>Voice agents are easy to demo and hard to scale. The architecture we ended up with fires every possible response in parallel and lets a fast router pick the winner mid-flight. This post walks through everything we tried before it, and why this design is the one we kept.</em></p> <hr> <h2 id="the-problem-nobodys-demo-solves">The problem nobody&rsquo;s demo solves</h2> <p>Every voice agent demo looks the same: clean audio, a cooperative user, a short scripted exchange. Then you ship it, and two things break immediately.</p> https://thiyaga.xyz/posts/reliable-voice-agents/ Wed, 29 Jul 2026 00:45:00 +0530 https://thiyaga.xyz/posts/reliable-voice-agents/ <p>Voice agent demos are easy to build. A worker, an STT → LLM → TTS pipeline, and you have something that talks. The hard part is making it reliable.</p> <p>We run AI call agents in production (ours are on LiveKit, but nothing below depends on that). A call lasts up to an hour. A lost recording is a compliance problem. A lost transcript means the user redoes the call. A silent agent means a human sitting in an empty room, wondering if anyone is there.</p>