<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Jottings of Vishal - agents</title>
  <link href="https://jottings.vishalvshekkar.com/tags/agents.html" rel="alternate" />
  <link href="https://jottings.vishalvshekkar.com/tags/agents/atom.xml" rel="self" type="application/atom+xml" />
  <id>https://jottings.vishalvshekkar.com/tags/agents/</id>
  <updated>2026-09-04T18:33:47.934Z</updated>
  <subtitle>Posts tagged with "agents"</subtitle>
  
  <entry>
    <title>Balancing cache writes-vs-utilization, tool calling back-and-forth, reasoning effort, to improve ...</title>
    <link href="https://jottings.vishalvshekkar.com/jots/5722626097.html" rel="alternate" />
    <id>https://jottings.vishalvshekkar.com/jots/5722626097.html</id>
    <updated>2026-09-04T18:33:46.260Z</updated>
    <summary>Balancing cache writes-vs-utilization, tool calling back-and-forth, reasoning effort, to improve subjective &amp; perceived sense of performance of an LLM-backed agentic workflow, while having a classifier to route each inference call to a complex hierarchy &amp; classes of LLMs, providers, and wrappers which also reroutes during outages, degraded outputs, and latencies, while ensuring a consistent personality to the end consuming entity is the funnest engineering challenge today.</summary>
  </entry>
  
</feed>