docs: add bounded MXL production pilot offer

docs/REVIEW-RESPONSE.md

@@ -0,0 +1,58 @@
+# Review response thread
+
+Shared async thread between the author and the independent reviewer. Each item carries a
+status: `open` / `fixed` / `disputed` / `wontfix` / `needs-decision` /
+`needs-hw-verify`.
+
+## Oct 9 — Guy's decisions + actions on the Oct 9 review
+
+Thanks for the thorough pass. Guy made the five calls; here they are, with what's done.
+
+### Decisions from Guy
+
+1. **Merge order** — `fixed(plan)`: proceed in the recommended order **#38 → #40 → #37 → #39 → #41**.
+   #37 is held until its open notes below are addressed; the other four can go once conflicts
+   are resolved.
+2. **Deploying unmerged branches to the live box** — `accepted`: continue only to close the
+   verification gap faster, with the merge plan above ending the drift. Box-specific details
+   stay uncommitted or placeholdered, never on master.
+3. **/join policy** — `decided`: **passphrase (merge #40)**. Public contributors must present
+   the SRT passphrase; the /join QR embeds it so legitimate scanners stay one-tap.
+4. **Apache copyright holder (#39)** — `decided`: **Guy Cochran** (individual).
+5. **Naming the hostname + Jonas** — `decided`: remove the live hostname from the repo; drop
+   the employer tag on the EFA changelog line; retain the public open-source proxy credit.
+
+### Actions already taken (Oct 9)
+
+- **Hostname leak (#41)** — `fixed`: committed documentation no longer hardcodes the live
+  server hostname.
+- **Control API hardening** — `fixed`: mediamtx's API binds to localhost rather than all
+  interfaces.
+- **Employer tag (#37 changelog)** — `fixed`: the entry states the technical cause without
+  identifying an individual or employer.
+
+### Still open — #37 notes
+
+- `open`: NMOS routing still ignores the flow ID; the domain ID is not a proper UUID; the ADR
+  over-claims both as done; the soak check reports a stale result as a failure; the PR is too
+  broad. Address these before merge.
+
+## Oct 9 (later) — two more decisions from Guy
+
+- **Multiview visibility (#40)** — `decided`: passphrase-to-publish is enough; public viewing
+  is acceptable for the demo.
+- **Core-vs-Skin leak sweep** — `fixed` in PR #42: runnable Core defaults are neutral while
+  the project publication may retain its authorship and field notes.
+
+## Oct 17 — commercialization validation page
+
+- **Pilot offer page** — `open`: `docs/pilot.html` adds a bounded, measured pilot offer and
+  explicitly excludes a general-purpose production-platform promise, a 24/7 SLA, universal
+  MXL-vs-NDI claims, and finished TAMS editing/replay.
+- **Claim discipline** — `fixed`: the page carries only already documented proof points and
+  labels the CPU comparison as a bounded same-host test.
+- **Runtime impact** — `fixed`: documentation-only; no Core, live-system, tunnel, cloud, Docker,
+  meeting, mapping or service changes.
+- **Review request** — `open`: verify that the public wording does not overstate the evidence
+  and that the pilot scope remains commercially useful without implying product maturity.
+

docs/index.html

@@ -74,6 +74,7 @@
       <a href="#architecture">Architecture</a>
       <a href="/what-is-mxl.html">What is MXL</a>
       <a href="/findings.html">Findings</a>
+      <a href="/pilot.html">Pilot</a>
       <a href="/learn.html">Learn MXL</a>
       <a href="#field">The field</a>
     </nav>
@@ -88,6 +89,7 @@ <h1>What if the switcher<br>was <em>shared memory?</em></h1>
         <p class="lede">A working exploration of a live production control room built around the Media eXchange Layer: real cameras, remote guests, playout, layouts, graphics, audio, multiview, recording and standard control — assembled as software on general-purpose cloud VMs.</p>
         <div class="hero-actions">
           <a class="button primary" href="/mxl-archive.html">See it in action <span aria-hidden="true">→</span></a>
+          <a class="button secondary" href="/pilot.html">Evaluate MXL with us</a>
           <a class="button secondary" href="https://github.com/guycochran/mxl-cloud-production-demo">Explore the source <span aria-hidden="true">↗</span></a>
           <a class="button secondary" href="QUICKSTART.md">Run it in ~10 minutes</a>
         </div>
@@ -326,9 +328,9 @@ <h2>From fresh Ubuntu VM to a cuttable MXL production in about ten minutes.</h2>
     <section class="closing">
       <div class="shell closing-inner">
         <p class="eyebrow">AN OPEN QUESTION, RUNNING AS CODE</p>
-        <h2>What should an MXL-native<br>switcher become?</h2>
-        <p>Fork it. Reproduce it. Challenge the measurements. Replace a function. Bring a different controller. File what breaks. The point is not to define the answer alone — it is to make the conversation testable.</p>
-        <div class="hero-actions"><a class="button primary" href="https://github.com/guycochran/mxl-cloud-production-demo">Join the work on GitHub <span aria-hidden="true">↗</span></a><a class="button secondary" href="https://github.com/guycochran/mxl-cloud-production-demo/issues">Open an issue</a></div>
+        <h2>What should MXL prove<br>for your team?</h2>
+        <p>Run the open reference stack, challenge the measurements, or define a bounded production pilot around one real workflow. The next step is not more features—it is a decision supported by working evidence.</p>
+        <div class="hero-actions"><a class="button primary" href="/pilot.html">Define a paid pilot <span aria-hidden="true">→</span></a><a class="button secondary" href="https://github.com/guycochran/mxl-cloud-production-demo">Join the work on GitHub <span aria-hidden="true">↗</span></a><a class="button secondary" href="https://github.com/guycochran/mxl-cloud-production-demo/issues">Open an issue</a></div>
       </div>
     </section>
   </main>