diff --git a/content/blog/post/2025-07-07-sea.md b/content/blog/post/2025-07-07-sea.md
new file mode 100644
index 0000000..8bf8b6a
--- /dev/null
+++ b/content/blog/post/2025-07-07-sea.md
@@ -0,0 +1,39 @@
+---
+type: post
+title: Introducing Search Engine Adapter
+author: Andrew Wray
+subtitle: One URL for Google, ChatGPT, and Wikipedia
+summary: SEA routes search queries to different services with nginx.
+date: 2025-07-07
+tags: ["sea", "nginx", "search"]
+---
+
+Our [previous post](/blog/post/2025-06-01-chatgpt-default-search-engine/)
+explained how to make ChatGPT your search engine.
+
+
+[Search Engine Adapter (SEA)](https://github.com/dense-analysis/sea)
+builds on that idea.
+
+SEA is a lightweight nginx proxy that chooses between Google, the ChatGPT API,
+or Wikipedia based on keywords in your query.
+It compiles nginx config files from a simple TOML source using Go, resulting in
+a small and secure deployment.
+
+## Key Features
+
+* Dynamic keyword routing with nginx's `map` directive.
+* One HTTPS endpoint easily secured with Let's Encrypt.
+* Minimal dependencies and small attack surface.
+
+You can configure SEA using `config-example.toml`, then compile the final
+`nginx.conf`. Once deployed, every search is forwarded to the matching service.
+
+A live instance is available at
+[https://search.denseanalysis.org/?q=%s](https://search.denseanalysis.org/?q=who+is+w0rp+and+what+is+ALE)
+for anyone who wants to try it. The instance does not log any access or error
+logs, so Dense Analysis pledges not to spy on your searches, but don't trust us.
+The configuration and program to make it are available on
+[GitHub](https://github.com/dense-analysis/sea), and you may deploy your own
+instance locally, remotely, or wherever you want. We welcome feedback and
+contributions!