vs-claude: '30 engineers' not 'pods'; plain-language card (1 on hanzo.ai out-built 30 on claude.ai)

defense/hanzo-vs-claude.tex

@@ -120,29 +120,28 @@ The platform's own engineering audit covered 334 repositories and
 history and normalizing cost against each contributor's monthly spend. We report
 the three metrics the audit treats as defensible: total commits shipped, unit cost
 per line of code (\$/LoC), and shipping rhythm per dollar of monthly spend
-(commits per \$1{,}000). The contractor pods are anonymized as Team~A, Team~B, and
-Team~C.
+(commits per \$1{,}000). The three contractor teams---\textbf{30 engineers} in
+total---are anonymized as Team~A, Team~B, and Team~C.
 
 \begin{table}[H]
 \centering
 \small
 \begin{tabular}{l r r r}
 \toprule
-\textbf{Group (same frontier model)} & \textbf{Commits} & \textbf{\$ / LoC} & \textbf{Commits / \$1k} \\
+\textbf{Who (all on the same model)} & \textbf{Commits} & \textbf{\$ / LoC} & \textbf{Commits / \$1k} \\
 \midrule
-One engineer on the Hanzo agentic stack & 15{,}294 & \$0.0006 & 588 \\
+\textbf{1 engineer} on Hanzo (\texttt{hanzo.ai}) & \textbf{15{,}294} & \textbf{\$0.0006} & \textbf{588} \\
 \midrule
-Team~A (contractor pod, conventional tooling) & 2{,}112 & \$0.179 & 105 \\
-Team~B (contractor pod, conventional tooling) & 498   & \$0.560 & 10  \\
-Team~C (contractor pod, conventional tooling) & 1{,}415 & \$0.153 & 32  \\
+Team~A --- 4 engineers (\texttt{claude.ai}) & 2{,}112 & \$0.179 & 105 \\
+Team~B --- 6 engineers (\texttt{claude.ai}) & 498   & \$0.560 & 10  \\
+Team~C --- 20 engineers (\texttt{claude.ai}) & 1{,}415 & \$0.153 & 32  \\
 \bottomrule
 \end{tabular}
-\caption{The client's own engineering audit, anonymized. All contributors used
-the same frontier chat model; only the single engineer's work was multiplied by
-Hanzo's agentic stack. By the audit's own unit-cost metric, that engineer's cost
-per line of code was roughly $250\times$ to $900\times$ lower than the contractor
-pods', and that engineer alone shipped more commits than the three contractor
-pods combined.}
+\caption{The client's own engineering audit, anonymized. Everyone used the same
+frontier model (Claude); only the single engineer also had Hanzo's agentic stack.
+That one engineer shipped more commits than all \textbf{30 engineers combined}, at a
+cost per line of code roughly $250\times$ to $900\times$ lower. One person on
+\texttt{hanzo.ai} out-built thirty people on \texttt{claude.ai}.}
 \end{table}
 
 \paragraph{Reading the numbers honestly.} The audit is explicit that the raw
@@ -315,8 +314,8 @@ approximately 97\% after migrating to this stack.
 The productivity comparison in Section~\ref{sec:oneeng} is also a cost argument. Hanzo's founding
 CEO/CTO, Zach Kelling, was---by public usage tracking (\texttt{claudecount.com})---the
 \textbf{\#1 user of Anthropic by token count worldwide in 2025}; the audited output
-behind that throughput, one engineer outproducing entire contractor pods on the
-same frontier model, is documented above. The point here is what changed
+behind that throughput---one engineer outproducing 30 engineers on the
+same frontier model---is documented above. The point here is what changed
 afterward: that same agentic throughput now runs on Hanzo's \emph{owned},
 self-hostable stack rather than a metered cloud API. The capability that topped a
 closed provider's global usage is exactly the capability Hanzo packages as a

site/src/config/papers.ts

@@ -68,7 +68,7 @@ export const siteConfig: SiteConfig = {
       id: 'hanzo-vs-claude',
       title: 'hanzo.ai vs. claude.ai',
       subtitle: 'Owned, self-hostable AI versus a centralized cloud chat',
-      abstract: 'Owned, self-hostable AI versus a metered cloud chat — on deployment and control, not benchmark scores. Includes a measured productivity case: on a regulated build where every engineer had the same frontier model (Claude), the client\'s own audit found one engineer on Hanzo\'s agentic stack shipped more than three contractor pods combined at ~250x-900x lower cost per line — the differentiator being the owned agentic stack, not the model. Covers data residency, air-gapped operation, provenance, post-quantum security, and cost at scale.',
+      abstract: 'The headline: on a regulated build where every engineer had the same model (Claude), one engineer using hanzo.ai out-shipped 30 engineers on claude.ai combined — at ~250x-900x lower cost per line. Same model; the difference was the agentic stack. Plus the deeper case for owned, self-hostable AI: data residency, air-gapped operation, post-quantum security, and cost at scale.',
       pdfUrl: '/pdfs/hanzo-vs-claude.pdf',
       latexUrl: 'https://github.com/hanzoai/papers/blob/main/defense/hanzo-vs-claude.tex',
       githubUrl: 'https://github.com/hanzoai/papers',