<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Engineering</title>
    <link>https://www.c1.ai/engineering/</link>
    <description>Recent content in Engineering on ConductorOne</description>
    <language>en-us</language>
    <lastBuildDate>Fri, 13 Mar 2026 07:00:00 GMT</lastBuildDate>
    <atom:link href="https://www.c1.ai/engineering/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>How We Cut Our Embedding Costs in Half and Shrank Storage by 50%</title>
      <link>https://www.c1.ai/engineering/how-we-cut-our-embedding-costs-in-half-and-shrank-storage-by-50/</link>
      <pubDate>Fri, 13 Mar 2026 07:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/how-we-cut-our-embedding-costs-in-half-and-shrank-storage-by-50/</guid>
      <description>How C1 refactored its embedding pipeline to achieve ~50% reduction in Bedrock inference costs, 50% reduction in vector storage, and embeddings that refresh in hours instead of days.</description>
    </item>
    <item>
      <title>Fault Injection for Complete Branch Coverage</title>
      <link>https://www.c1.ai/engineering/fault-injection-for-complete-branch-coverage/</link>
      <pubDate>Fri, 20 Feb 2026 02:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/fault-injection-for-complete-branch-coverage/</guid>
      <description>Error handling branches are invisible to valid inputs. Fault injection makes them visible.</description>
    </item>
    <item>
      <title>Proving Equivalence with E-Graphs</title>
      <link>https://www.c1.ai/engineering/proving-equivalence-with-egraphs/</link>
      <pubDate>Fri, 20 Feb 2026 01:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/proving-equivalence-with-egraphs/</guid>
      <description>Two permission derivations can look completely different and compute the same access. E-graphs find out.</description>
    </item>
    <item>
      <title>Every Branch Condition Compiles to a DFA</title>
      <link>https://www.c1.ai/engineering/every-branch-condition-compiles-to-a-dfa/</link>
      <pubDate>Fri, 20 Feb 2026 00:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/every-branch-condition-compiles-to-a-dfa/</guid>
      <description>Every branch condition in every connector compiles to a deterministic finite automaton. Here&apos;s how that enables static branch prediction.</description>
    </item>
    <item>
      <title>One Mock Server, Twelve Protocols</title>
      <link>https://www.c1.ai/engineering/one-mock-server-twelve-protocols/</link>
      <pubDate>Thu, 12 Feb 2026 00:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/one-mock-server-twelve-protocols/</guid>
      <description>A YAML-driven mock server that speaks twelve protocols through one interface. Adding a new connector means writing a config file. </description>
    </item>
    <item>
      <title>Formally Verifying Two Hundred Identity Connectors</title>
      <link>https://www.c1.ai/engineering/formally-verifying-identity-connectors/</link>
      <pubDate>Wed, 11 Feb 2026 00:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/formally-verifying-identity-connectors/</guid>
      <description>Connector verification reduces to three set operations. Here&apos;s the framework that proves two hundred identity connectors correct.</description>
    </item>
    <item>
      <title>Six Shapes of Authorization</title>
      <link>https://www.c1.ai/engineering/six-shapes-of-authorization/</link>
      <pubDate>Tue, 10 Feb 2026 13:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/six-shapes-of-authorization/</guid>
      <description>We axiomatized over 1,800 permission systems. Every one of them reduces to one of six shapes.</description>
    </item>
    <item>
      <title>Splitting AI Agents to Contain Prompt Injection</title>
      <link>https://www.c1.ai/engineering/splitting-ai-agents-to-contain-prompt-injection/</link>
      <pubDate>Thu, 05 Jun 2025 13:00:00 GMT</pubDate>
      <guid>https://www.c1.ai/engineering/splitting-ai-agents-to-contain-prompt-injection/</guid>
      <description>Go behind the scenes of how we built guardrails into Thomas, C1&apos;s AI agent—and the industry&apos;s first multi-agent identity security platform. This blog breaks down our architecture, trust models, and sandboxing techniques that ensure agents can automate identity governance safely, securely, and transparently.</description>
    </item>
  </channel>
</rss>