diff --git a/.gitignore b/.gitignore
index 7d20525..29047dd 100644
--- a/.gitignore
+++ b/.gitignore
@@ -65,3 +65,4 @@ certs/
 memory-bank/epoch/*
 !memory-bank/epoch/.gitkeep
 models/*.bin
+whisper.cpp/
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 9a45f06..4429534 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,12 @@ All notable changes to OpenStudio will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [Unreleased]
+
+### Added
+
+- **`setup-ai.sh`** one-shot bootstrap script for the AI pipeline. It clones and builds `whisper.cpp`, downloads the default Whisper model into `models/`, and prompts for LLM provider settings in `.env`.
+
 ## [0.3.2] - 2026-05-25
 
 ### Added
diff --git a/QUICKSTART.md b/QUICKSTART.md
index 6efa626..d4cecf4 100644
--- a/QUICKSTART.md
+++ b/QUICKSTART.md
@@ -9,6 +9,8 @@ cp .env.example .env
 cd server && npm install && cd ..
 cd web && npm install && cd ..
 cp station-manifest.sample.json station-manifest.json
+# Optional: bootstrap the AI pipeline (whisper.cpp, model, LLM config)
+./setup-ai.sh
 ```
 
 ## Daily Development Workflow
diff --git a/README.md b/README.md
index 001330c..048f23c 100644
--- a/README.md
+++ b/README.md
@@ -132,18 +132,18 @@ OpenStudio includes a complete post-production pipeline — transcription, show
 
 ### Prerequisites
 - `ffmpeg` and `ffprobe` on your `PATH` (most package managers ship both: `brew install ffmpeg`, `apt install ffmpeg`)
+- `cmake` on your `PATH` for the `whisper.cpp` build
 - ~1.5 GB free disk space for the default Whisper model
 
-### whisper.cpp setup (one-time)
+### One-shot AI setup
 
 ```bash
-git clone https://github.com/ggerganov/whisper.cpp
-cd whisper.cpp && make -j$(nproc)
-cd .. && mkdir -p models
-wget -O models/ggml-medium.bin https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-medium.bin
+./setup-ai.sh
 ```
 
-You can swap the model — `ggml-tiny.bin` (~75 MB) is faster but less accurate; `ggml-large.bin` is the other direction. The server looks for `models/ggml-medium.bin` by default.
+The script clones and builds `whisper.cpp`, downloads `models/ggml-medium.bin`, and prompts for your LLM provider so `.env` is ready for show notes.
+
+You can swap the model manually if you want — `ggml-tiny.bin` (~75 MB) is faster but less accurate; `ggml-large.bin` is the other direction. The server looks for `models/ggml-medium.bin` by default.
 
 ### LLM Provider Examples
 
@@ -240,8 +240,6 @@ See [docs/vision.md](docs/vision.md) for the full project vision and philosophy.
 Honest about what's there and what isn't:
 
 - **Invite-link UI** — the server can mint scoped invite tokens (host / ops / guest, 4 h TTL), but the host UI doesn't expose a button yet. For now, hosts share the room URL manually. Coming in 0.4.
-- **AI pipeline setup is manual** — the whisper.cpp build, model download, and LLM configuration aren't scripted. A `setup-ai.sh` would be a great PR.
-- **whisper.cpp gitlink** — the repo references whisper.cpp as a gitlink without a `.gitmodules` entry. Use the manual `git clone` in the AI setup section above; `git submodule update` will fail.
 - **Mesh scale ceiling** — WebRTC mesh tops out around 15 participants. Larger rooms need an SFU (planned for 0.5).
 
 ## Contributing
diff --git a/package.json b/package.json
index 5e7535e..13db3a3 100644
--- a/package.json
+++ b/package.json
@@ -7,6 +7,7 @@
     "start": "node server/server.js",
     "dev": "node --watch server/server.js",
     "test": "node server/test-signaling.js && node server/test-rooms.js",
+    "test:setup-ai": "bash tests/test-setup-ai.sh",
     "install": "cd server && npm install"
   },
   "engines": {
diff --git a/run-pre-validation.sh b/run-pre-validation.sh
index 979a204..78d5d3c 100755
--- a/run-pre-validation.sh
+++ b/run-pre-validation.sh
@@ -18,6 +18,7 @@ echo "  3. test-gain-controls.mjs   - Per-participant gain controls"
 echo "  4. test-program-bus.mjs     - Program bus mixing"
 echo "  5. test-mix-minus.mjs       - Mix-minus calculation (3 peers)"
 echo "  6. test-return-feed.mjs     - Return feed routing (2 peers)"
+echo "  7. test-setup-ai.sh         - AI bootstrap script smoke test"
 echo ""
 echo "================================================================================"
 echo ""
@@ -32,13 +33,18 @@ run_test() {
     local test_name=$1
     local test_file=$2
     local timeout_seconds=${3:-60}
+    local runner="node"
+
+    case "$test_file" in
+        *.sh) runner="bash" ;;
+    esac
 
     echo "--- Running: $test_name ---"
     echo "File: $test_file"
     echo "Timeout: ${timeout_seconds}s"
     echo ""
 
-    if timeout ${timeout_seconds} node "$test_file"; then
+    if timeout ${timeout_seconds} "$runner" "$test_file"; then
         echo ""
         echo "✅ PASSED: $test_name"
         echo ""
@@ -64,6 +70,7 @@ run_test "Gain Controls" "tests/test-gain-controls.mjs" 60
 run_test "Program Bus Mixing" "tests/test-program-bus.mjs" 60
 run_test "Mix-Minus Calculation" "tests/test-mix-minus.mjs" 60
 run_test "Return Feed Routing" "tests/test-return-feed.mjs" 60
+run_test "AI Bootstrap Script" "tests/test-setup-ai.sh" 120
 
 # Summary
 echo "================================================================================"
@@ -76,7 +83,7 @@ for test in "${TESTS[@]}"; do
 done
 
 echo ""
-echo "Results: $PASSED passed, $FAILED failed (out of 6 total)"
+echo "Results: $PASSED passed, $FAILED failed (out of 7 total)"
 echo ""
 
 # Decision
diff --git a/server/lib/whisper-transcriber.js b/server/lib/whisper-transcriber.js
index 546e28e..f94cff4 100644
--- a/server/lib/whisper-transcriber.js
+++ b/server/lib/whisper-transcriber.js
@@ -175,12 +175,41 @@ async function downloadModel(model = 'medium') {
  */
 async function buildWhisperCpp() {
   const whisperDir = path.join(process.cwd(), 'whisper.cpp');
+  const whisperBin = WHISPER_BIN;
+  const cmakeLists = path.join(whisperDir, 'CMakeLists.txt');
+  const makefile = path.join(whisperDir, 'Makefile');
+
+  if (!fs.existsSync(whisperDir)) {
+    logger.info('Cloning whisper.cpp...');
+    await execAsync(`git clone --depth 1 --recursive https://github.com/ggerganov/whisper.cpp.git ${whisperDir}`);
+  } else if (!fs.existsSync(cmakeLists) && !fs.existsSync(makefile)) {
+    const entries = fs.readdirSync(whisperDir);
+    if (entries.length === 0) {
+      logger.warn('Found empty whisper.cpp directory, re-cloning...');
+      fs.rmSync(whisperDir, { recursive: true, force: true });
+      await execAsync(`git clone --depth 1 --recursive https://github.com/ggerganov/whisper.cpp.git ${whisperDir}`);
+    } else {
+      throw new Error('whisper.cpp checkout exists but has no supported build files');
+    }
+  }
 
-  logger.info('Cloning whisper.cpp...');
-  await execAsync(`git clone --depth 1 https://github.com/ggerganov/whisper.cpp.git ${whisperDir}`);
+  if (fs.existsSync(whisperBin)) {
+    logger.info('whisper.cpp already built');
+    return;
+  }
 
-  logger.info('Building whisper.cpp...');
-  await execAsync(`cd ${whisperDir} && make -j$(nproc) 2>&1 | tail -5`);
+  if (fs.existsSync(cmakeLists)) {
+    logger.info('Configuring whisper.cpp with CMake...');
+    await execAsync(`cmake -S ${whisperDir} -B ${whisperDir}/build -DWHISPER_BUILD_TESTS=OFF -DWHISPER_BUILD_EXAMPLES=ON -DWHISPER_BUILD_SERVER=OFF`);
+
+    logger.info('Building whisper.cpp...');
+    await execAsync(`cmake --build ${whisperDir}/build --config Release --target whisper-cli -j$(nproc)`);
+  } else if (fs.existsSync(makefile)) {
+    logger.info('Building whisper.cpp...');
+    await execAsync(`cd ${whisperDir} && make -j$(nproc) 2>&1 | tail -5`);
+  } else {
+    throw new Error('whisper.cpp checkout does not contain a supported build system');
+  }
 
   logger.info('whisper.cpp built successfully');
 }
diff --git a/setup-ai.sh b/setup-ai.sh
new file mode 100755
index 0000000..c513751
--- /dev/null
+++ b/setup-ai.sh
@@ -0,0 +1,283 @@
+#!/bin/bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+cd "$ROOT_DIR"
+
+ENV_FILE=".env"
+WHISPER_DIR="whisper.cpp"
+WHISPER_BIN="$WHISPER_DIR/build/bin/whisper-cli"
+MODEL_DIR="models"
+MODEL_FILE="$MODEL_DIR/ggml-medium.bin"
+WHISPER_REPO="https://github.com/ggerganov/whisper.cpp.git"
+MODEL_URL="https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-medium.bin"
+
+info() { printf '%s\n' "$*"; }
+ok() { printf 'OK: %s\n' "$*"; }
+warn() { printf 'WARN: %s\n' "$*" >&2; }
+fail() { printf 'ERROR: %s\n' "$*" >&2; exit 1; }
+
+require_cmd() {
+  command -v "$1" >/dev/null 2>&1 || fail "Missing required command: $1"
+}
+
+cpu_jobs() {
+  if command -v nproc >/dev/null 2>&1; then
+    nproc
+  elif command -v sysctl >/dev/null 2>&1; then
+    sysctl -n hw.ncpu
+  else
+    echo 4
+  fi
+}
+
+download_file() {
+  local url="$1"
+  local output="$2"
+
+  if command -v curl >/dev/null 2>&1; then
+    curl -L --fail --silent --show-error "$url" -o "$output"
+  elif command -v wget >/dev/null 2>&1; then
+    wget -q "$url" -O "$output"
+  else
+    fail "Need curl or wget to download the Whisper model"
+  fi
+}
+
+ensure_env_file() {
+  if [ ! -f "$ENV_FILE" ]; then
+    [ -f .env.example ] || fail "Missing .env.example"
+    cp .env.example "$ENV_FILE"
+    ok "Created .env from .env.example"
+  fi
+}
+
+load_env() {
+  set -a
+  # shellcheck disable=SC1091
+  . "./$ENV_FILE"
+  set +a
+}
+
+set_env_value() {
+  local key="$1"
+  local value="$2"
+  local tmp
+  tmp="$(mktemp)"
+
+  awk -v key="$key" -v value="$value" '
+    BEGIN { found = 0 }
+    $0 ~ "^" key "=" { print key "=" value; found = 1; next }
+    { print }
+    END { if (!found) print key "=" value }
+  ' "$ENV_FILE" > "$tmp"
+
+  mv "$tmp" "$ENV_FILE"
+}
+
+prompt_default() {
+  local prompt="$1"
+  local default_value="$2"
+  local reply
+
+  printf '%s [%s]: ' "$prompt" "$default_value" >&2
+  read -r reply
+  printf '%s\n' "${reply:-$default_value}"
+}
+
+ensure_whisper_cpp() {
+  local has_build_files=0
+
+  if [ -f "$WHISPER_DIR/CMakeLists.txt" ] || [ -f "$WHISPER_DIR/Makefile" ]; then
+    has_build_files=1
+  fi
+
+  if [ -d "$WHISPER_DIR" ] && [ "$has_build_files" -eq 0 ]; then
+    if [ -z "$(find "$WHISPER_DIR" -mindepth 1 -maxdepth 1 -print -quit 2>/dev/null)" ]; then
+      warn "Found empty whisper.cpp/ directory; replacing it with a fresh clone"
+    else
+      warn "Found whisper.cpp/ without build files; replacing it with a fresh clone"
+    fi
+    rm -rf "$WHISPER_DIR"
+  fi
+
+  if [ ! -d "$WHISPER_DIR" ]; then
+    info "Cloning whisper.cpp..."
+    git clone --depth 1 --recursive "$WHISPER_REPO" "$WHISPER_DIR"
+  elif [ "$has_build_files" -eq 1 ]; then
+    ok "Found whisper.cpp"
+  fi
+
+  if [ -x "$WHISPER_BIN" ]; then
+    ok "whisper.cpp already built"
+    return
+  fi
+}
+
+build_whisper_cpp() {
+  if [ -x "$WHISPER_BIN" ]; then
+    ok "whisper.cpp already built"
+    return
+  fi
+
+  local jobs
+  jobs="$(cpu_jobs)"
+
+  if [ -f "$WHISPER_DIR/CMakeLists.txt" ]; then
+    require_cmd cmake
+    info "Configuring whisper.cpp with CMake..."
+    cmake -S "$WHISPER_DIR" -B "$WHISPER_DIR/build" \
+      -DWHISPER_BUILD_TESTS=OFF \
+      -DWHISPER_BUILD_EXAMPLES=ON \
+      -DWHISPER_BUILD_SERVER=OFF
+
+    info "Building whisper.cpp..."
+    cmake --build "$WHISPER_DIR/build" --config Release --target whisper-cli -j"$jobs"
+  elif [ -f "$WHISPER_DIR/Makefile" ]; then
+    require_cmd make
+    info "Building whisper.cpp..."
+    (cd "$WHISPER_DIR" && make -j"$jobs")
+  else
+    fail "whisper.cpp is missing supported build files"
+  fi
+
+  [ -x "$WHISPER_BIN" ] || fail "whisper.cpp build finished but $WHISPER_BIN is missing"
+  ok "Built whisper.cpp"
+}
+
+ensure_model() {
+  mkdir -p "$MODEL_DIR"
+
+  if [ -f "$MODEL_FILE" ]; then
+    ok "Whisper model already present"
+    return
+  fi
+
+  info "Downloading Whisper model..."
+  download_file "$MODEL_URL" "$MODEL_FILE"
+  [ -f "$MODEL_FILE" ] || fail "Model download failed"
+  ok "Downloaded Whisper model"
+}
+
+configure_llm() {
+  local current_base_url="${LLM_BASE_URL:-http://localhost:1234/v1}"
+  local current_model="${LLM_MODEL:-qwen3.5-35b}"
+  local current_api_key="${LLM_API_KEY:-}"
+  local choice
+  local provider_label
+  local base_url
+  local model
+  local api_key
+
+  info ""
+  info "Choose an LLM provider for show notes:"
+  info "  1) LM Studio (default)"
+  info "  2) Ollama"
+  info "  3) OpenAI"
+  info "  4) Together AI"
+  info "  5) Groq"
+  info "  6) Custom OpenAI-compatible / shim"
+  printf 'Provider [1]: '
+  read -r choice
+  choice="${choice:-1}"
+
+  case "$choice" in
+    1)
+      provider_label="LM Studio"
+      base_url="http://localhost:1234/v1"
+      model="qwen3.5-35b"
+      api_key=""
+      ;;
+    2)
+      provider_label="Ollama"
+      base_url="http://localhost:11434/v1"
+      model="llama3.3"
+      api_key=""
+      ;;
+    3)
+      provider_label="OpenAI"
+      base_url="https://api.openai.com/v1"
+      model="gpt-4o-mini"
+      printf 'Model [gpt-4o-mini]: '
+      read -r model
+      model="${model:-gpt-4o-mini}"
+      printf 'API key: '
+      read -r -s api_key
+      printf '\n'
+      ;;
+    4)
+      provider_label="Together AI"
+      base_url="https://api.together.xyz/v1"
+      model="meta-llama/Llama-3.3-70B-Instruct-Turbo"
+      printf 'Model [meta-llama/Llama-3.3-70B-Instruct-Turbo]: '
+      read -r model
+      model="${model:-meta-llama/Llama-3.3-70B-Instruct-Turbo}"
+      printf 'API key: '
+      read -r -s api_key
+      printf '\n'
+      ;;
+    5)
+      provider_label="Groq"
+      base_url="https://api.groq.com/openai/v1"
+      model="llama-3.3-70b-versatile"
+      printf 'Model [llama-3.3-70b-versatile]: '
+      read -r model
+      model="${model:-llama-3.3-70b-versatile}"
+      printf 'API key: '
+      read -r -s api_key
+      printf '\n'
+      ;;
+    6)
+      provider_label="Custom"
+      base_url="$(prompt_default 'Base URL' "$current_base_url")"
+      model="$(prompt_default 'Model' "$current_model")"
+      printf 'API key [leave blank if none]: '
+      read -r -s api_key
+      printf '\n'
+      api_key="${api_key:-$current_api_key}"
+      ;;
+    *)
+      warn "Unknown choice, using LM Studio defaults"
+      provider_label="LM Studio"
+      base_url="http://localhost:1234/v1"
+      model="qwen3.5-35b"
+      api_key=""
+      ;;
+  esac
+
+  set_env_value "LLM_BASE_URL" "$base_url"
+  set_env_value "LLM_MODEL" "$model"
+  set_env_value "LLM_API_KEY" "$api_key"
+
+  ok "Configured LLM provider: $provider_label"
+}
+
+main() {
+  require_cmd git
+  require_cmd awk
+
+  if ! command -v ffmpeg >/dev/null 2>&1; then
+    fail "ffmpeg is required. Install it first (brew install ffmpeg or apt install ffmpeg)."
+  fi
+
+  if ! command -v ffprobe >/dev/null 2>&1; then
+    fail "ffprobe is required. Install it with ffmpeg before continuing."
+  fi
+
+  ensure_env_file
+  load_env
+
+  ensure_whisper_cpp
+  build_whisper_cpp
+  ensure_model
+  configure_llm
+
+  info ""
+  ok "AI pipeline setup complete"
+  info "Next steps:"
+  info "  1) Start OpenStudio: npm start"
+  info "  2) Or run the dev workflow: ./dev.sh start"
+  info "  3) Reload the studio UI so /api/capabilities picks up the new tools"
+}
+
+main "$@"
diff --git a/tests/test-setup-ai.sh b/tests/test-setup-ai.sh
new file mode 100755
index 0000000..85d2b59
--- /dev/null
+++ b/tests/test-setup-ai.sh
@@ -0,0 +1,97 @@
+#!/bin/bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+TMPDIR="$(mktemp -d)"
+trap 'rm -rf "$TMPDIR"' EXIT
+
+cp "$ROOT_DIR/setup-ai.sh" "$TMPDIR/setup-ai.sh"
+cp "$ROOT_DIR/.env.example" "$TMPDIR/.env.example"
+mkdir -p "$TMPDIR/fakebin" "$TMPDIR/fake-whisper-src"
+: > "$TMPDIR/fake-whisper-src/CMakeLists.txt"
+
+cat > "$TMPDIR/fakebin/git" <<'EOF'
+#!/bin/bash
+set -euo pipefail
+
+if [ "${1:-}" = "clone" ]; then
+  target="${@: -1}"
+  src="${FAKE_WHISPER_SRC:?}"
+  rm -rf "$target"
+  mkdir -p "$target"
+  cp -R "$src"/. "$target"/
+  exit 0
+fi
+
+command git "$@"
+EOF
+
+cat > "$TMPDIR/fakebin/cmake" <<'EOF'
+#!/bin/bash
+set -euo pipefail
+
+if [ "${1:-}" = "-S" ]; then
+  builddir=""
+  while [ "$#" -gt 0 ]; do
+    case "$1" in
+      -B) builddir="$2"; shift 2 ;;
+      *) shift ;;
+    esac
+  done
+
+  mkdir -p "$builddir/bin"
+  cat > "$builddir/bin/whisper-cli" <<'EOB'
+#!/bin/sh
+exit 0
+EOB
+  chmod +x "$builddir/bin/whisper-cli"
+  exit 0
+fi
+
+exit 0
+EOF
+
+cat > "$TMPDIR/fakebin/curl" <<'EOF'
+#!/bin/bash
+set -euo pipefail
+
+out=""
+while [ "$#" -gt 0 ]; do
+  case "$1" in
+    -o) out="$2"; shift 2 ;;
+    *) shift ;;
+  esac
+done
+
+[ -n "$out" ] || exit 1
+mkdir -p "$(dirname "$out")"
+printf 'model' > "$out"
+EOF
+
+cat > "$TMPDIR/fakebin/ffmpeg" <<'EOF'
+#!/bin/sh
+exit 0
+EOF
+
+cat > "$TMPDIR/fakebin/ffprobe" <<'EOF'
+#!/bin/sh
+exit 0
+EOF
+
+chmod +x "$TMPDIR/fakebin/"*
+
+(
+  cd "$TMPDIR"
+  PATH="$TMPDIR/fakebin:$PATH" FAKE_WHISPER_SRC="$TMPDIR/fake-whisper-src" \
+    bash ./setup-ai.sh <<'EOF'
+2
+EOF
+)
+
+grep -q '^LLM_BASE_URL=http://localhost:11434/v1$' "$TMPDIR/.env"
+grep -q '^LLM_MODEL=llama3.3$' "$TMPDIR/.env"
+grep -q '^LLM_API_KEY=$' "$TMPDIR/.env"
+[ -x "$TMPDIR/whisper.cpp/build/bin/whisper-cli" ]
+[ -f "$TMPDIR/models/ggml-medium.bin" ]
+
+echo "setup-ai.sh test passed"
diff --git a/web/js/main.js b/web/js/main.js
index 42aa811..a50de02 100644
--- a/web/js/main.js
+++ b/web/js/main.js
@@ -283,8 +283,7 @@ class OpenStudioApp {
       out.push({
         name: 'whisper.cpp',
         present: whisperPresent,
-        install:
-          'git clone https://github.com/ggerganov/whisper.cpp\ncd whisper.cpp && make -j$(nproc)',
+        install: './setup-ai.sh   # clones whisper.cpp and builds whisper-cli',
       });
 
       const modelPresent = !!(caps.model && caps.model.available);
@@ -292,8 +291,7 @@ class OpenStudioApp {
       out.push({
         name: `Whisper model${sizeHint}`,
         present: modelPresent,
-        install:
-          'mkdir -p models\nwget -O models/ggml-medium.bin \\\n  https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-medium.bin',
+        install: './setup-ai.sh   # downloads the medium Whisper model into models/',
       });
     }
 
@@ -304,8 +302,7 @@ class OpenStudioApp {
       out.push({
         name: `LLM polish (optional) — ${caps.llm.provider || 'OpenAI-compatible'}`,
         present: llmReady,
-        install:
-          '# In .env:\nLLM_BASE_URL=http://localhost:1234/v1   # LM Studio default\nLLM_MODEL=qwen3.5-35b\n# Show notes still generate without an LLM — they just fall back\n# to a transcript-derived title and summary.',
+        install: './setup-ai.sh   # choose an LLM provider and write it into .env',
       });
     }
 
diff --git a/whisper.cpp b/whisper.cpp
index 46ca43d..1fe009c 160000
--- a/whisper.cpp
+++ b/whisper.cpp
@@ -1 +1 @@
-Subproject commit 46ca43d6399fdeada1b49fb2126ba373bd9ebc38
+Subproject commit 1fe009caeda75f69bc864d6370b10674e45a92bd