<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Waveguide Blog</title><description>Engineering and product notes from the Waveguide team: agent security, honest metering, missions, and what it takes to trust an AI with real accounts.</description><link>https://waveguide.adwave.com/</link><item><title>Missions: giving an agent a goal instead of a task</title><link>https://waveguide.adwave.com/blog/missions-goals-not-tasks/</link><guid isPermaLink="true">https://waveguide.adwave.com/blog/missions-goals-not-tasks/</guid><description>Task-based AI does what you say and stops. A mission is a goal your agent pursues for weeks — research, outreach, replies, booked meetings — inside rails that make autonomy safe.</description><pubDate>Fri, 03 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Honest metering: multi-model routing and the 80/100/120 budget ladder</title><link>https://waveguide.adwave.com/blog/honest-metering/</link><guid isPermaLink="true">https://waveguide.adwave.com/blog/honest-metering/</guid><description>How an always-on agent stays affordable: route every job to the cheapest model that does it well, suspend the machine between wakes, and bound the worst-case bill by construction.</description><pubDate>Mon, 29 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Approvals by construction: how T0–T3 risk tiers work</title><link>https://waveguide.adwave.com/blog/approvals-by-construction/</link><guid isPermaLink="true">https://waveguide.adwave.com/blog/approvals-by-construction/</guid><description>Every action your agent takes is classified into one of four risk tiers, and the dangerous ones physically require a human. Here&apos;s the taxonomy and why it&apos;s enforced in the network layer.</description><pubDate>Mon, 22 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Why the model never holds a credential</title><link>https://waveguide.adwave.com/blog/model-never-holds-a-credential/</link><guid isPermaLink="true">https://waveguide.adwave.com/blog/model-never-holds-a-credential/</guid><description>The single design decision that makes Waveguide trustable with real accounts: OAuth tokens live in a vault, and a proxy injects them after the request leaves the model.</description><pubDate>Mon, 15 Jun 2026 00:00:00 GMT</pubDate></item></channel></rss>