Option 5: Siri Voice-First Integration

---

What you're building

Sample interactions: - 🎤 "Hey Siri, morning brief" → 🔊 "You have 3 items due this week. Ethics assignment in 4 days..." - 🎤 "Hey Siri, tracker add" → 🔊 "What to add?" → 🎤 "GPU quiz Oct 15" → 🔊 "Added row 8" - 🎤 (in car, hands-free) "Hey Siri, what's due tomorrow" → 🔊 spoken response

---

Prerequisite

Complete Option 3 (Apps Script + Shortcut) first - Siri voice sits on top of that infrastructure. If you've done Option 3, you're 80% of the way here.

---

Setup Overview

---

1Verify Option 3 works

Before adding voice, confirm the underlying flow works:

• Your Apps Script Web App URL responds to POST requests
• Your Claude API key is valid
• You have a basic Shortcut that can add rows

If not, complete OPTION_3_APPS_SCRIPT_SHORTCUT.md first.

---

2Build the voice-first Shortcut

This time, redesign the shortcut around voice input and voice output.

2.1 - Create new shortcut

Shortcuts app → + → name it "Tracker Voice"

2.2 - Add actions in order

Action 1: Dictate Text (voice input)

• Search: Dictate Text
• Language: English (India) or your locale
• Stop Listening: On Tap or After Pause (short pause = 1.5s)

Action 2: Get Contents of URL (call Claude)

Same as Option 3 Step 4, but modify system prompt for voice:

• URL: https://api.anthropic.com/v1/messages
• Method: POST
• Headers: (same as Option 3)
• Body: json { "model": "claude-sonnet-4-5", "max_tokens": 512, "system": "You parse spoken tracker requests. Voice input may have transcription errors — be forgiving. Valid subjects: FoVL-Language, GPU Programming, Ethics and Professional Life, Innovation and IP Management, Advanced AI. Return ONLY JSON: {action, subject, type, title, due_date (DD-Mmm-YYYY), time (HH:MM IST), notes}. Include 'voice_reply' field with a natural-sounding sentence I should speak back to the user.", "messages": [ {"role": "user", "content": [DICTATED_TEXT_VARIABLE]} ] }

Action 3: Extract voice_reply

• Get Dictionary Value → get content[0].text
• Then Get Dictionary Value → get voice_reply

Action 4: Speak Text

• Text: the voice_reply from Action 3
• Language: same as Dictate
• Rate: 0.5 (natural pace)
• Pitch: 1.0
• Wait Until Finished: On

Action 5: POST to Apps Script (for writes)

Only run if action != "read": - Wrap Actions 5-6 in an If block: If action is not "read"

Inside the If: - Get Contents of URL → POST parsed JSON to your Apps Script URL

Action 6: Optional final Speak (confirmation)

• Speak: "Done." or the response from Apps Script

---

3Add to Siri

1. Open the Tracker Voice shortcut
2. Tap the settings icon (top right, ⓘ)
3. Add to Siri → tap the red record button
4. Say your trigger phrase clearly: "tracker" or "log tracker" or "morning brief"
5. Tap Done

Trigger phrases that work well:

• "Hey Siri, tracker check" (read)
• "Hey Siri, tracker add" (write)
• "Hey Siri, morning brief" (daily summary)
• "Hey Siri, what's due" (upcoming)

Trigger phrases to avoid:

• Anything with "sheet" (Siri may launch Google Sheets app instead)
• Anything one word (may conflict with other shortcuts)
• Anything with numbers or dates in the phrase itself

---

4Test on iPhone

1. Lock your phone
2. Say "Hey Siri, tracker check"
3. Speak your query when Siri activates the shortcut: "What's due this week"
4. Listen for the spoken reply

Common test cases: - "What's due tomorrow?" → should list items - "Add GPU quiz October 15 at 6 PM" → should ask confirmation, then add - "Mark row 5 as done" → status update

---

5Enable Apple Watch

1. Shortcuts app → tap Tracker Voice
2. Settings icon → toggle Show on Apple Watch ON
3. On Apple Watch → open Shortcuts app → your shortcut appears

Or set it as a Watch complication: 1. Long-press watch face → Edit → Complications 2. Choose slot → select Shortcuts → pick Tracker Voice 3. Now it's one-tap from your wrist

Voice on Watch works via Raise to Speak - no button press needed if configured.

---

6Test CarPlay + AirPods

CarPlay

1. Connect to your car
2. Long-press Siri button on steering wheel
3. Say: "Hey Siri, morning brief"
4. Reply plays through car speakers

AirPods

• Squeeze/tap AirPod → "Tracker check"
• Reply plays in your ear

HomePod / HomePod mini

• "Hey Siri, tracker check" from anywhere in room
• Reply plays through HomePod

---

Step 7 (Optional): Multiple specialized shortcuts

Instead of one big shortcut, create several focused ones:

Each does one thing well = better voice recognition + faster response.

---

Advanced ideas

Automations (no voice, triggered by events)

Shortcuts app → Automation tab → +

Examples: - When I arrive at work → speak "Today's tracker items:" - 9 AM every weekday → read new items overnight - When I open Google Sheets app → auto-run "Show updated items" - When I connect AirPods → speak morning brief

Focus Mode integration

Only enable tracker voice during work hours: - Settings → Focus → Work → Add Shortcut trigger

Multiple language support

• Set Dictate Text language to Hindi, Tamil, etc.
• Claude Sonnet 4.5 handles most languages natively
• Voice reply also speaks in same language

---

Pros / Cons

---

Sample voice_reply examples

Good system prompts produce natural spoken responses:

User says: "What's due this week?" Voice reply: "You have 3 items this week. Ethics assignment in 4 days, GPU quiz on Saturday, and Innovation assignment next Monday."

User says: "Add GPU quiz October 15 at 6 PM" Voice reply: "Adding GPU Programming quiz on October 15th at 6 PM IST. Say yes to confirm."

User says: "Nothing due" Voice reply: "You're all caught up. Nothing due in the next 7 days."

---

Troubleshooting

Siri doesn't hear "tracker" correctly

• Re-record the phrase more slowly
• Try a longer phrase: "check my tracker" instead of just "tracker"
• Ensure Siri language matches your accent (Settings → Siri & Search → Language)

Voice reply cut off mid-sentence

• Speak action: enable Wait Until Finished
• Increase Claude's max_tokens if replies are truncated

Works on iPhone but not Apple Watch

• Watch needs the shortcut synced: iPhone Shortcuts app → toggle "Show on Apple Watch"
• If it uses SSH, add the SSH key to Watch keychain separately

CarPlay says "I can't help with that"

• Enable Use with CarPlay in shortcut settings
• CarPlay's Siri is stricter - use simpler trigger phrases

Dictation not stopping automatically

• Change Stop Listening to After Pause (2 seconds)
• Or use After Tap if pauses interrupt your thinking

---

Cost note

Voice adds no extra cost - same Claude API calls as text. Voice recognition is on-device (Siri) or included in iOS.

---

Summary

You now have hands-free tracker access from any Apple device. Best combined with Option 3 (the infrastructure) or Option 4 (for batch text access).

The full stack: - Mac at desk: Claude Code direct - iPhone away from Mac: SSH (Option 1) or Shortcut (Option 3) - Voice / hands-free: Siri (Option 5) - Batch access: WhatsApp bot (Option 4) or Telegram bot (Option 2)

All layered on the same Google Sheet as source of truth. 🚀