<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Zeeshan Ahmad Lone — AI Engineer</title><description>Software engineer focused on conversational AI, ML systems and production-grade voice experiences — from real-time audio pipelines to agents that hold useful conversations.</description><link>https://zeeshanlone.com/</link><item><title>Designing prompts for outbound conversations</title><link>https://zeeshanlone.com/blog/designing-prompts-for-outbound-conversations/</link><guid isPermaLink="true">https://zeeshanlone.com/blog/designing-prompts-for-outbound-conversations/</guid><description>Outbound calls invert the usual agent contract: the agent wants something and the human didn&apos;t ask for the call. That changes how you write the prompt.</description><pubDate>Mon, 22 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Twenty milliseconds: the unit that shapes realtime audio</title><link>https://zeeshanlone.com/blog/twenty-milliseconds/</link><guid isPermaLink="true">https://zeeshanlone.com/blog/twenty-milliseconds/</guid><description>Everything in a realtime voice pipeline happens in 20ms frames. Understanding that one number explains most of the architecture around it.</description><pubDate>Tue, 14 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Why voice agents fail at turn-taking</title><link>https://zeeshanlone.com/blog/why-voice-agents-fail-at-turn-taking/</link><guid isPermaLink="true">https://zeeshanlone.com/blog/why-voice-agents-fail-at-turn-taking/</guid><description>Most voice agents don&apos;t fail at understanding — they fail at knowing when to speak. A look at endpointing, barge-in, and the awkward silence problem.</description><pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate></item></channel></rss>