<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Gravel · Field Notes</title>
    <link>https://gravel.work/field-notes/</link>
    <atom:link href="https://gravel.work/rss.xml" rel="self" type="application/rss+xml" />
    <description>Dispatches on the forward deployed motion: deployment economics, org design, and what enterprise clients grade you on.</description>
    <language>en</language>
    <item>
      <title>The true cost of a forward deployed motion</title>
      <link>https://gravel.work/field-notes/true-cost-of-a-forward-deployed-motion/</link>
      <guid isPermaLink="true">https://gravel.work/field-notes/true-cost-of-a-forward-deployed-motion/</guid>
      <pubDate>Mon, 20 Jul 2026 00:00:00 GMT</pubDate>
      <description>A forward deployed engineer costs $280–400K fully loaded: a $173,816 median base, grossed up roughly 31% for benefits, plus travel, tooling, and management overhead. At about two concurrent engagements per person, field cost lands at $125–200K per engagement per year, which is the effective ACV floor for white-glove deployment. Move engagements-per-person to six and that floor drops well below $100K.</description>
      <category>Deployment economics</category>
      <category>Unit economics</category>
      <category>Forward deployed engineering</category>
    </item>
    <item>
      <title>What your client is actually grading you on</title>
      <link>https://gravel.work/field-notes/what-clients-expect-from-deployment/</link>
      <guid isPermaLink="true">https://gravel.work/field-notes/what-clients-expect-from-deployment/</guid>
      <pubDate>Tue, 16 Jun 2026 00:00:00 GMT</pubDate>
      <description>Enterprise clients grade a deployment on two axes they rarely state out loud: speed to value and depth of fit. Every question they ask — &quot;what does week one look like?&quot;, &quot;who&apos;s our point of contact?&quot;, &quot;how custom is this?&quot; — is one of those two in disguise. The five standard failures are all boundary failures: asks, status, decisions, and evidence falling into the strip between two companies that no tool owns.</description>
      <category>Client expectations</category>
      <category>Deployment</category>
      <category>Enterprise sales</category>
    </item>
    <item>
      <title>Building a forward deployed org when agents do half the job</title>
      <link>https://gravel.work/field-notes/building-a-forward-deployed-org/</link>
      <guid isPermaLink="true">https://gravel.work/field-notes/building-a-forward-deployed-org/</guid>
      <pubDate>Mon, 11 May 2026 00:00:00 GMT</pubDate>
      <description>The forward deployed engineer title covers three different jobs: builder FDE (60% of postings), sales engineer with a commit bit (30%), and internal tools builder (10%). Roughly half of the real role is coordination — status updates, chasing approvals, meeting-note archaeology — and that half is what agentic tooling now carries. The org-design consequence is to hire fewer, more senior people and change the unit of capacity from accounts carried to engagements supervised.</description>
      <category>Hiring</category>
      <category>Forward deployed engineering</category>
      <category>Org design</category>
    </item>
    <item>
      <title>How forward deployment actually works: Palantir, read closely</title>
      <link>https://gravel.work/field-notes/palantir-forward-deployment-playbook/</link>
      <guid isPermaLink="true">https://gravel.work/field-notes/palantir-forward-deployment-playbook/</guid>
      <pubDate>Thu, 02 Apr 2026 00:00:00 GMT</pubDate>
      <description>Palantir split engineering in two: forward deployed engineers who lived at customer sites and were licensed to overfit, and product development engineers who generalized those field hacks into platform primitives. The loop — embed, solve, generalize, resell — is what turned a services cost structure into 80% gross margins, because each pass made the next deployment cheaper. Teams that copy the embedded-engineer half without the generalization half end up with unmaintainable bespoke deployments at Accenture margins.</description>
      <category>Palantir</category>
      <category>Forward deployed engineering</category>
      <category>Product strategy</category>
    </item>
    <item>
      <title>The deal dies between the demo and the workflow</title>
      <link>https://gravel.work/field-notes/deal-dies-between-demo-and-workflow/</link>
      <guid isPermaLink="true">https://gravel.work/field-notes/deal-dies-between-demo-and-workflow/</guid>
      <pubDate>Thu, 26 Feb 2026 00:00:00 GMT</pubDate>
      <description>Enterprise AI deals die in the six months after the demo — in data access requests, security reviews, and the gap between what the product does and how the buyer&apos;s team actually works. The market&apos;s answer is the forward deployed engineer, but at $280–400K fully loaded across roughly two accounts, the motion costs $125–200K per engagement per year. A large share of that week is coordination rather than engineering, and coordination is the class of work agents can now carry.</description>
      <category>Forward deployed engineering</category>
      <category>Enterprise AI</category>
      <category>Deployment economics</category>
    </item>
  </channel>
</rss>
