diff --git a/index.html b/index.html
index 7e7ba05..c65db25 100644
--- a/index.html
+++ b/index.html
@@ -1181,7 +1181,7 @@ async function askBreak(kind) {
     messages: [{ role: "user", content: `You are the DJ of g*ndora (say it "gandora"), a radio station that writes its own songs for one listener.
 Seed: "${state.seed}"${state.station ? `, station "${state.station}"` : ""}.${state.djName ? ` Your DJ name is ${state.djName}.` : ""}${state.djPersona ? ` Your persona, and the way you speak: ${state.djPersona}.` : ""} The last tracks: ${prev}.
 Listener likes: ${state.taste.like.join("; ") || "(unknown)"}. Avoids: ${state.taste.avoid.join("; ") || "(unknown)"}. Their words: ${state.taste.notes.join(" | ") || "(none)"}.
-Write a 30-second INTERMISSION in the station's own sound: ${BREAK_BRIEF[kind]}
+Write a 30-second INTERMISSION in the station's own sound: ${state.djName ? BREAK_BRIEF[kind].replace(/Pick a DJ name[^)]*\) and open/, "Open") : BREAK_BRIEF[kind]}
 ${spoken ? "It is SPOKEN by a voice, not sung: lyrics are exactly what the DJ says, 4-7 short lines of natural speech, no rhyming, no section markers, no stage directions. tags describe the station's sound (for the record), bpm/key/energy match the station." : "4-8 lines of lyrics."}
 Return JSON: title (short${kind === "welcome" ? "; for the welcome, the title is your DJ name" : ""}), tags (the music prompt), lyrics, bpm, key, energy, seconds (30), meter, station ("${state.station || state.seed}"), line (one line the DJ says on screen, under 70 characters)` }] };
   const d = await venice("/chat/completions", body);