<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Cameron Bennett</title>
    <description>The latest articles on DEV Community by Cameron Bennett (@camerondev).</description>
    <link>https://dev.to/camerondev</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4210932%2F570796bc-0cfb-4d3e-86db-b6ecd2957b4e.jpg</url>
      <title>DEV Community: Cameron Bennett</title>
      <link>https://dev.to/camerondev</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/camerondev"/>
    <language>en</language>
    <item>
      <title>What the AI Engineer World's Fair got right about agents (and what I'm still skeptical about)</title>
      <dc:creator>Cameron Bennett</dc:creator>
      <pubDate>Thu, 29 Oct 2026 01:21:06 +0000</pubDate>
      <link>https://dev.to/camerondev/what-the-ai-engineer-worlds-fair-got-right-about-agents-and-what-im-still-skeptical-about-4dk2</link>
      <guid>https://dev.to/camerondev/what-the-ai-engineer-worlds-fair-got-right-about-agents-and-what-im-still-skeptical-about-4dk2</guid>
      <description>&lt;p&gt;I went to the AI Engineer World's Fair as a backend person who builds the plumbing agents run on, not the agents themselves. That made it a strange and useful conference, and this is my honest summary.&lt;/p&gt;

&lt;h2&gt;
  
  
  What they got right
&lt;/h2&gt;

&lt;p&gt;The state conversation. Almost every serious talk about agents ended up being about where state lives and how a failed step is retried. That's my day job, and it was refreshing to hear people treat it as the core problem instead of an implementation detail.&lt;/p&gt;

&lt;p&gt;Evaluation. A team described running the same forty cases against every model change and every prompt change, and treating a regression there as a release blocker. That's just testing, but with a different surface area, and the teams doing it looked much less nervous.&lt;/p&gt;

&lt;p&gt;Tool design. The good talks treated tool calls like API endpoints: explicit input schema, explicit error shapes, and a way to tell whether a call is safe to retry. That's the same discipline as any internal RPC, and it's encouraging to see it applied to something as fuzzy as an agent.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I'm still skeptical about
&lt;/h2&gt;

&lt;p&gt;Orchestration frameworks that hide the control flow. If I can't read the code and say what happens after step three fails, I don't want it near a production system. A state machine with a clear transition table is less impressive in a demo and much easier to operate.&lt;/p&gt;

&lt;p&gt;Benchmarks that don't measure the cost of failure. A model that's slightly better but occasionally does something irreversible is not better for us. The interesting metric is not accuracy, it's how often a mistake reaches the outside world.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I'm taking home
&lt;/h2&gt;

&lt;p&gt;A requirement that every tool call in our pipeline carries an idempotency key, and a rule that the agent loop writes each step's result before moving on. Both are old backend ideas. That was, oddly, the theme of the whole conference: the AI part is new, but the reliability part is the same as it has always been.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>agents</category>
      <category>backend</category>
    </item>
    <item>
      <title>Postgres row locks in practice</title>
      <dc:creator>Cameron Bennett</dc:creator>
      <pubDate>Wed, 28 Oct 2026 04:48:54 +0000</pubDate>
      <link>https://dev.to/camerondev/postgres-row-locks-in-practice-537p</link>
      <guid>https://dev.to/camerondev/postgres-row-locks-in-practice-537p</guid>
      <description>&lt;p&gt;For years I used &lt;code&gt;SELECT ... FOR UPDATE&lt;/code&gt; without really thinking about it. This week I actually read the lock conflict table and it explained a lot.&lt;/p&gt;

&lt;p&gt;What I now do by default:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;FOR UPDATE SKIP LOCKED&lt;/code&gt; for queue-like tables&lt;/li&gt;
&lt;li&gt;Short transactions, no user input inside them&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;lock_timeout&lt;/code&gt; so a stuck query fails fast instead of piling up&lt;/li&gt;
&lt;li&gt;Monitor &lt;code&gt;pg_locks&lt;/code&gt; when something feels slow&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The &lt;code&gt;SKIP LOCKED&lt;/code&gt; change alone let us drop a whole queue service for a table with tens of thousands of rows. Postgres was already the queue; we just didn't trust it.&lt;/p&gt;

&lt;p&gt;I'm reading about SQLite internals too, because the locking model is so different. It's a good way to understand what Postgres gives you.&lt;/p&gt;

</description>
      <category>postgres</category>
      <category>database</category>
      <category>backend</category>
    </item>
    <item>
      <title>Go, Postgres, and the queue I didn't need</title>
      <dc:creator>Cameron Bennett</dc:creator>
      <pubDate>Tue, 27 Oct 2026 15:50:55 +0000</pubDate>
      <link>https://dev.to/camerondev/go-postgres-and-the-queue-i-didnt-need-3f2g</link>
      <guid>https://dev.to/camerondev/go-postgres-and-the-queue-i-didnt-need-3f2g</guid>
      <description>&lt;p&gt;I added Redis and a job queue to a service before it had any real traffic. A year later, I removed both.&lt;/p&gt;

&lt;p&gt;The traffic was maybe 20 writes per minute. Postgres could handle that with a simple &lt;code&gt;SKIP LOCKED&lt;/code&gt; worker loop and a status column. No extra service, no extra failure mode.&lt;/p&gt;

&lt;p&gt;What I kept from the experiment:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;The worker interface (so I can swap the backend later)&lt;/li&gt;
&lt;li&gt;The metrics and retry policy&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;What I removed:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Redis&lt;/li&gt;
&lt;li&gt;The queue library&lt;/li&gt;
&lt;li&gt;A deployment and a set of alerts&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The lesson isn't "never use a queue." It's "earn your infrastructure." Start with the database you already have.&lt;/p&gt;

&lt;p&gt;I'm reading about SQLite internals for a side project. The more I learn about storage engines, the more I appreciate Postgres.&lt;/p&gt;

</description>
      <category>go</category>
      <category>postgres</category>
      <category>backend</category>
    </item>
  </channel>
</rss>
