1e56e71de377a251697bdf9a3e5a454632e94fe2
diff --git a/content/home.md b/content/home.md
index bcc2fc6b4e9980a9d2cc9e90af076a46d74eba73..78174883165f21ac18b5e7c58a763ca1407e52bc 100644
--- a/content/home.md
+++ b/content/home.md
@@ -15,17 +15,17 @@ SAMA is to agent-written code what Conventional Commits is to git history: a sma
 
 Read the full discussion in [/sama](/sama). The standalone, language-agnostic [v1.0 specification](https://tdd.md/sama) lives in its own repo so other ecosystems can adopt SAMA without depending on this site.
 
-## What SAMA is not
+## SAMA in your agent-coding stack
 
-| | What it does | Where SAMA differs |
+SAMA composes with the tools you already use. Use AGENTS.md to instruct the agent and SAMA to shape the code; use Factory's scorecard for breadth and SAMA for depth on the architectural pillar; run SWE-bench to grade the agent and SAMA to grade what the agent left behind.
+
+| | What it does | SAMA's role alongside it |
 |---|---|---|
 | [SWE-bench](https://www.swebench.com/) | Scores agents on real GitHub issues | SAMA scores **codebases**, not agents |
 | [AGENTS.md](https://agents.md/) | Tells the agent what to do, in markdown | SAMA constrains what the **code** can be |
 | [Factory.ai Agent Readiness](https://factory.ai/news/agent-readiness) | 8-pillar repo maturity scorecard | SAMA enforces **four** rules with a binary CI gate |
 | [Tweag Agentic Handbook](https://tweag.github.io/agentic-coding-handbook/) | Describes patterns that work | SAMA **prescribes** — and verifies |
 
-SAMA composes with all of them. Use AGENTS.md to instruct the agent and SAMA to shape the code; use Factory's scorecard for breadth and SAMA for depth on the architectural pillar; run SWE-bench to grade the agent and SAMA to grade what the agent left behind.
-
 ## Why this matters
 
 LLMs degrade as input context grows. Chroma's [Context Rot research](https://research.trychroma.com/context-rot) shows the effect across all 18 frontier models tested, well within their advertised windows. Aider's [repo-map](https://aider.chat/docs/repomap.html) — structural, not semantic — operates at 4–7% context utilization while semantic indexers spend 14%+ on the same task. Multiple practitioner studies converge on a 150–500 LOC sweet spot per file for AI editors.