diff --git a/changelog/1.1.0.md b/changelog/1.1.0.md
new file mode 100644
index 0000000..8a1bd86
--- /dev/null
+++ b/changelog/1.1.0.md
@@ -0,0 +1,38 @@
+# 1.1.0 — 2026-06-30
+
+This release turns BrainCue from a live read-along copilot into a full interview
+**trainer**: prep grounded materials before the call, rehearse out loud with a voice
+interviewer, and answer with proof — every cue tied back to your own background.
+
+## Added
+
+- **Grounded, proof-linked answers.** Live answers now cite their sources inline — each
+  claim drawn from your résumé/JD carries a numbered marker, and the Cue Card renders
+  those as tappable **source chips** that expand to the exact text. A new **fabrication
+  guard** means that when your background can't support what's asked, BrainCue says so and
+  pivots to a transferable-skills framing instead of inventing experience.
+
+- **Pre-Interview Brief.** One click on any interview generates a tailored prep brief from
+  your résumé × the job description × company research: the **likely questions** (ranked),
+  your **coverage gaps** (what the role wants that your résumé is thin on, plus how to
+  bridge each), **strengths to lead with**, and **company angles** — all grounded.
+
+- **STAR Story Bank.** Extract a bank of reusable **STAR stories** from your résumé, tagged
+  by competency (leadership, conflict, impact, …). Browse, edit, regenerate, and delete
+  them on your profile — and they automatically **ground your live answers**.
+
+- **Sparring — a two-way voice mock.** The AI interviewer asks a question **aloud**, you
+  **answer by speaking** (hold to talk), and each answer gets a rated, specific critique —
+  strengths, what to work on, and a tip pointing to a story you could have used — before it
+  asks a follow-up. A real back-and-forth rehearsal.
+
+- **"Story to tell" live cue.** During a live interview, the single best-matching story from
+  your bank surfaces in the Cue Card as a **📖 Story to tell** prompt for the current
+  question — your ready-made answer, right when you need it.
+
+## Under the hood
+
+- Every feature is grounded-only (no invented experience) and defensively parsed, and each
+  was adversarially reviewed before merge. Story regeneration is atomic and embeds before it
+  writes, so a hiccup can never wipe your curated bank; push-to-talk voice capture is
+  hardened against mic-stream leaks and stuck turns. The unit-test suite grows to 102.
diff --git a/package.json b/package.json
index d972dac..7f81a33 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "ai-interview-assistant",
-  "version": "1.0.0",
+  "version": "1.1.0",
   "description": "BrainCue Copilot — desktop AI interview copilot (Electron + React + OpenAI). Local-first data, BYO OpenAI key.",
   "author": "tpikachu",
   "license": "MIT",