<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Use Cases on SAM</title><link>https://sam-mesh.dev/docs/use-cases/</link><description>Recent content in Use Cases on SAM</description><generator>Hugo</generator><language>en-us</language><atom:link href="https://sam-mesh.dev/docs/use-cases/index.xml" rel="self" type="application/rss+xml"/><item><title>Warm Agent Pool</title><link>https://sam-mesh.dev/docs/use-cases/warm-agent-pool/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://sam-mesh.dev/docs/use-cases/warm-agent-pool/</guid><description>&lt;p>Fan a batch of work across a pool of identical, already-running worker agents —
built entirely from ordinary mesh MCP services, no gossip and no node changes.&lt;/p>
&lt;p>Source: &lt;a href="https://github.com/google/sam/tree/main/development/examples/code-reviewer-pool">&lt;code>development/examples/code-reviewer-pool/&lt;/code>&lt;/a>.&lt;/p>
&lt;h2 id="the-idea">The idea&lt;/h2>
&lt;p>Some agent tools are expensive to stand up but cheap to reuse — a reviewer that
shells out to an LLM, a sandbox that boots a runtime, a service holding a warm
model in memory. You don&amp;rsquo;t want to spawn one per request, and you don&amp;rsquo;t want a
single instance serialising all your work. What you want is a &lt;strong>pool of
identical, already-running workers&lt;/strong> and something that hands them out one job at
a time.&lt;/p></description></item><item><title>Gemini Buddy</title><link>https://sam-mesh.dev/docs/use-cases/gemini-buddy/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://sam-mesh.dev/docs/use-cases/gemini-buddy/</guid><description>&lt;p>Hold a real multi-turn conversation with a &lt;em>second&lt;/em> LLM exposed as an ordinary
mesh service — where your agent never resends the conversation, because the
service remembers its own side.&lt;/p>
&lt;p>Source: &lt;a href="https://github.com/google/sam/tree/main/development/examples/gemini-buddy-mcp">&lt;code>development/examples/gemini-buddy-mcp/&lt;/code>&lt;/a>.&lt;/p>
&lt;h2 id="the-idea">The idea&lt;/h2>
&lt;p>Sometimes you want your agent to &lt;em>talk to another model&lt;/em> — a second opinion, a
peer to argue a design with, a specialist that builds up its own context over
many turns. The naive way is to carry both conversations in the orchestrator:
your agent would have to store the other model&amp;rsquo;s transcript and replay all of it
on every turn.&lt;/p></description></item></channel></rss>