Polish Fide AI organization card

README.md

@@ -8,29 +8,32 @@ short_description: Open research and benchmarks for faith-aware AI evaluation.
 
 # Fide AI
 
-Fide AI develops open research, benchmarks, and public standards for evaluating AI systems in faith, moral guidance, and pastoral-care-adjacent contexts.
+**Open research and benchmarks for faith-aware AI evaluation.**
 
-Our work focuses on:
+Fide AI develops public benchmarks, evaluation methods, and responsible-use
+standards for AI systems used around faith, moral guidance, and pastoral-care
+adjacent questions.
 
-- theological triage and tradition-aware evaluation
-- faith and moral guidance benchmarks
-- structured harness design for safer AI behavior
-- responsible-use standards for spiritually sensitive AI contexts
+## Research Focus
 
-## Projects
+- Theological triage and tradition-aware evaluation
+- Faith and moral guidance benchmarks
+- Structured harness design for safer model behavior
+- Responsible-use standards for spiritually sensitive AI contexts
 
-### FMG-Bench
+## Featured Work
 
-[FMG-Bench](https://huggingface.co/datasets/FideAI/fmg-bench) is the Faith & Moral Guidance Benchmark for evaluating large language models in theological triage and pastoral guidance contexts.
-
-FMG-Bench tests whether models can distinguish between primary doctrine, secondary disagreement, lower-certainty theological questions, and pastoral application scenarios where safety, agency, and human referral matter.
+| Project | Description |
+|---|---|
+| [FMG-Bench](https://huggingface.co/datasets/FideAI/fmg-bench) | Faith & Moral Guidance Benchmark for evaluating large language models in theological triage and pastoral guidance contexts. |
+| [GitHub repo](https://github.com/FideAI/fmg-bench) | Dataset, runner, scoring spec, papers, and reproducibility materials. |
 
 ## Links
 
-- Website: https://fideai.org
-- GitHub: https://github.com/FideAI
-- FMG-Bench GitHub: https://github.com/FideAI/fmg-bench
+[Website](https://fideai.org) · [GitHub](https://github.com/FideAI) · [FMG-Bench Dataset](https://huggingface.co/datasets/FideAI/fmg-bench)
 
 ## Responsible Use
 
-Fide AI research is intended to make AI behavior in spiritually sensitive contexts visible, comparable, and reviewable. Our benchmarks do not certify pastoral authority, clinical safety, legal adequacy, or deployment readiness.
+Fide AI benchmarks make AI behavior in spiritually sensitive contexts visible,
+comparable, and reviewable. They do not certify pastoral authority, clinical
+safety, legal adequacy, denominational standing, or deployment readiness.

index.html

@@ -5,17 +5,67 @@
   <meta name="viewport" content="width=device-width, initial-scale=1">
   <title>Fide AI</title>
   <style>
-    body { font-family: system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif; margin: 3rem auto; max-width: 760px; line-height: 1.6; padding: 0 1.25rem; color: #172033; }
-    a { color: #164ea6; }
+    :root { color-scheme: light dark; }
+    body {
+      margin: 0;
+      font-family: Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
+      background: transparent;
+      color: inherit;
+    }
+    .wrap {
+      max-width: 880px;
+      padding: 28px 32px;
+    }
+    h1 {
+      margin: 0 0 8px;
+      font-size: 32px;
+      line-height: 1.1;
+      letter-spacing: 0;
+    }
+    .lede {
+      margin: 0 0 18px;
+      font-size: 18px;
+      font-weight: 650;
+    }
+    p {
+      margin: 0 0 16px;
+      max-width: 760px;
+      line-height: 1.55;
+    }
+    .grid {
+      display: grid;
+      grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
+      gap: 10px;
+      margin: 20px 0;
+    }
+    .item {
+      border: 1px solid rgba(127, 127, 127, 0.25);
+      border-radius: 8px;
+      padding: 12px 14px;
+    }
+    .item strong { display: block; margin-bottom: 3px; }
+    .links { display: flex; flex-wrap: wrap; gap: 10px; margin-top: 18px; }
+    a { color: #2563eb; text-decoration: none; font-weight: 650; }
+    a:hover { text-decoration: underline; }
+    .note { margin-top: 18px; font-size: 14px; opacity: 0.8; }
   </style>
 </head>
 <body>
-  <h1>Fide AI</h1>
-  <p>Open research and benchmarks for faith-aware AI evaluation.</p>
-  <ul>
-    <li><a href="https://huggingface.co/datasets/FideAI/fmg-bench">FMG-Bench dataset</a></li>
-    <li><a href="https://github.com/FideAI/fmg-bench">FMG-Bench GitHub</a></li>
-    <li><a href="https://fideai.org">Fide AI website</a></li>
-  </ul>
+  <main class="wrap">
+    <h1>Fide AI</h1>
+    <p class="lede">Open research and benchmarks for faith-aware AI evaluation.</p>
+    <p>Fide AI develops public benchmarks, evaluation methods, and responsible-use standards for AI systems used around faith, moral guidance, and pastoral-care adjacent questions.</p>
+    <section class="grid" aria-label="Research focus">
+      <div class="item"><strong>Theological triage</strong><span>Evaluating whether systems answer the right kind of faith question in the right way.</span></div>
+      <div class="item"><strong>FMG-Bench</strong><span>A benchmark for theological triage and pastoral guidance behavior in large language models.</span></div>
+      <div class="item"><strong>Structured harnesses</strong><span>Instruction-layer design for safer grounding, preference handling, and escalation boundaries.</span></div>
+    </section>
+    <div class="links">
+      <a href="https://huggingface.co/datasets/FideAI/fmg-bench">FMG-Bench Dataset</a>
+      <a href="https://github.com/FideAI/fmg-bench">GitHub Repo</a>
+      <a href="https://fideai.org">Website</a>
+    </div>
+    <p class="note">Benchmarks make behavior visible and reviewable. They do not certify pastoral authority, clinical safety, legal adequacy, or deployment readiness.</p>
+  </main>
 </body>
 </html>