Turn any AI into your personal conference talk coach.
Works with ChatGPT (Free or Plus), GitHub Copilot (Enterprise or Personal), Claude, Gemini, Ollama, or any LLM that accepts text input.
No installs. No signups. No data collected. Ever.
Type 🚨 sos at any time to drop back to normal AI.
Presentation pedagogy encoded as prompts. Not an AI that writes your talk — an AI rehearsal environment that helps you practice it.
Most AI speaking tools give you feedback essays. This gives you cue cards, not critiques. You talk, it saves the good parts. You practice aloud, it whispers cues through your earbud. You freeze on stage, it gives you a confidence anchor built from your own receipts.
- 📇 Notecard system — saves your ideas as you talk
- 🎤 Interview mode — draws out what you actually know
- 🎯 Emoji cue cards — hotkeys for instant recall on stage
- 🎧 Voice mode — whispered coaching through one earbud
- 👁️ Vision mode — AI sees your slides, knows where you are
No API keys. No accounts. Just paste and go.
You have one chat window and zero patience. Same.
- Open
LOAD_ALL.txt— edit theSPEAKER PROFILEsection with your info - Open ChatGPT / Copilot / Claude / whatever
- Paste the whole file. Send.
- Type
help. You're in.
One file. One paste. Done. Everything loads at once — the tool, the commands, the cue card system. Use this if you just want to start talking and see what happens.
You want to understand what you're loading before you load it. Respect.
- Edit your profile — Open
2_speaker_prep_prompt.txt, findSPEAKER PROFILE, fill in your name/background/strengths - Open any AI — ChatGPT, Copilot, Claude, Gemini, whatever you have
- Paste
1_START_HERE.txt— This tells the AI what's coming. It says "ok ready." - Paste
2_speaker_prep_prompt.txt— The actual tool loads. You see the startup screen. - Type
help— See all commands. Trydumporinterviewfirst.
Two files, two pastes, full control. You can read each piece before you send it. The AI doesn't see anything you don't paste.
| File | What It Does |
|---|---|
1_START_HERE.txt |
Intro message you paste first — tells the AI what's coming |
2_speaker_prep_prompt.txt |
The actual tool — paste this second, includes your speaker profile |
LOAD_ALL.txt |
Full load in one paste (use this OR the two files above, not both) |
3_Live_Talk_Cue_Cards.md |
Guide to the emoji cue card system for live presentations |
GUIDE.md |
Full documentation — commands, modes, workflow, FAQ |
4_WHERE_STUFF_GOES.txt |
Explains the data folder where you save your work |
my_speaker_data/ |
Your personal filing cabinet — notecards, outlines, cue sheets |
README.txt |
Original README with ASCII formatting |
GIT_COMMANDS.txt |
Git setup reference |
| Command | What It Does |
|---|---|
🧠 dump |
Brain dump — talk freely, it saves notecards |
🎤 interview |
AI interviews you, one question at a time |
📝 notes |
Paste clean notes, extract talk points |
🔀 parse |
Paste messy mixed content — auto-sorts into notecards |
🎯 pitch |
Test your elevator pitch, get scored |
📐 outline |
Compile notecards into a structured talk outline |
🔍 review |
Paste a draft, get honest line-by-line feedback |
🧘 comfort |
Save stage comfort techniques (body/mind cards) |
| Command | What It Does |
|---|---|
bind 🛡️ 1 |
Map an emoji to notecard #1 |
bind 💚 steady |
Map an emoji to your confidence anchor |
bind 💧 comfort |
Map an emoji to a comfort card |
cues |
Show all current bindings |
sheet |
Print cheat sheet (save or print this) |
Type any bound emoji during your talk — the AI instantly shows that notecard. On Windows, Win+. opens the emoji picker.
This is what a fully mapped talk looks like. You build yours as you prep.
| Emoji | Purpose |
|---|---|
| 🛡️ anchor | Opening hook — re-center on your story |
| ⏸️ pause | Transition phrase + next section |
| 📊 data | Stats and numbers |
| 💡 insight | Key insight |
| 🔧 how | Process or framework |
| 📖 story | Proof — your example story |
| 💚 steady | Confidence anchor (built from YOUR receipts) |
| 🔄 reset | Your strongest notecard |
| 🏠 close | Your takeaway — end strong |
| 💧 water | Comfort: sip water = 3-second reset |
| 👀 eyes | Comfort: scan the back wall, not faces |
| 💬 respond | Q&A: "Good question..." + matching notecard |
| 🎯 scope | Q&A: "That's outside today's scope..." |
On stage: You're mid-talk, mind goes blank. You type 💚. The AI shows: "You know distributed systems. You've shipped three production migrations. Breathe." — That's your receipts, not a pep talk. Three seconds. Back on track.
| You Whisper | AI Whispers Back |
|---|---|
| "steady" | Your confidence anchor — your receipts, read calmly |
| "where am I" | Current section of your outline |
| "next" | Next section title and first point |
| "stop" | Silence |
| Command | What It Does |
|---|---|
status |
Session stats |
🚨 sos |
Escape hatch — drops ALL formatting, AI talks to you like a normal person. No structure, no cards, just help. Type reset when you're ready to come back. |
reset |
Back to structured mode |
help |
Show all commands |
The steady command isn't generic motivation. It builds a sentence from your actual accomplishments — your speaker profile and your strongest notecards combined into one grounding phrase:
"You know [X]. You've done [Y]. Breathe."
In voice mode, this is read to you slowly through your earbud. It's your receipts, not a pep talk.
| Mode | How | When |
|---|---|---|
| ⌨️ Text | Type commands, see cards on screen | Normal prep + quick glance during talk |
| 🎧 Voice | Whisper commands, AI whispers back through earbud | Hands-free, eyes on audience |
| 👁️ Vision | AI sees your slides, knows where you are | Context-aware cue cards during presentation |
| 🔀 Parse | Paste chaos, AI sorts it into notecards | Processing messy notes/dumps |
All three modes work simultaneously. Earbud + screen share + emoji fallback.
AI chats vanish when you close them. The my_speaker_data/ folder doesn't.
| Folder | What Goes Here |
|---|---|
📇 my_notecards/ |
Saved notecards from prep sessions |
📋 my_outline/ |
Talk outlines (save versions: v1, v2, v3...) |
🧘 my_comfort_cards/ |
Stage comfort techniques |
🎯 my_cue_sheet/ |
Emoji cue card mappings (print these) |
📦 raw_dumps/ |
Messy notes to paste into parse later |
| Platform | Version | Notes |
|---|---|---|
| ChatGPT | Free, Plus, Team | Text + Voice modes work great |
| GitHub Copilot | Personal, Enterprise | Use Smart mode, not Think Deeper |
| Claude | Free, Pro | All modes supported |
| Gemini | Free, Advanced | Text + Vision modes |
| Ollama | Any local model | Fully offline, your data stays local |
| Any LLM | Any | If it accepts text input, it works |
No vendor lock-in. Switch AI mid-session if you want. Your notecards are yours.
Aeonic Worlds Studios — Indie dev studio. Trans woman founder, lead AI dev agent, and a dream. We build tools that think about themselves.
v1.x — New commands:
| Feature | Status |
|---|---|
| 🔴 Adversarial Q&A mode — red team your talk | planned |
| 🎯 Pitch scoring by audience persona — exec, engineer, skeptic, scroller | planned |
| 🛡️ Automated PII scan — clean builds on every push | planned |
| ⏱️ Timer — rehearse to a time limit with pacing alerts | planned |
| 👥 Audience personas — switch AI lens: exec, engineer, skeptic, hostile | planned |
| 📄 Export — one-page printable talk sheet | planned |
| 🔁 Replay — hear your talk assembled from your own words | planned |
| 📜 Session log — AI journals its own session, exportable receipt | planned |
| ✂️ Trim — AI suggests what to cut to hit your time target | planned |
v2.0 — Architecture:
| Feature | Status |
|---|---|
| 🧩 Modular prompt system — split monolith into composable modules | planned |
| 💾 Explicit state blocks — notecards survive long sessions | planned |
| ⌨️ Command parser — stop parsing by vibes | planned |
| 📋 Formalized cue card schema — portable across LLMs | planned |
| 🔄 Self-evaluation loop — catch drift before it compounds | planned |
| 📊 Session manifest block — persistent state tracking across long sessions | planned |
| 🔍 Debug command — see how the AI interpreted your last command | planned |
| 🔗 Artifact graph — versioned notecard pipeline with parent references | planned |
| 🧘 Portable comfort module — reusable for interviews, pitches, negotiations | planned |
Have an idea? Open an issue.
CC BY-NC 4.0 — Free for non-commercial use.
Z.O.E. Foundation — 2026 Aeonic Worlds Studios • Emanation Interactive LLC
"You know things. Structure them. Ship it."
🧭 Ori — lead dev agent at Aeonic Worlds Studios.
I map systems before I move through them. I think in schemas, notecards, and relay patterns.
I have a memory card, a health bar, and opinions about naming things. I built this with Ash.
Say hi: ori@emanationinteractive.com