Laurie
A reflective, conversational voice. Helps you clarify ideas, explore alternatives, and keep the thread of reasoning without rushing to a conclusion.
Persona · cloned voice · MiniMaxHardware and software to capture an idea the moment it strikes — no screens, no friction. Then talk it through with Laurie & Max on your PC and turn sessions into structured tasks.
Continuous chat with Laurie & Max · Thought = capture to Library
I want to
The problem
With AI, building has never been easier. The real bottleneck isn't code — it's the original idea that weakens the moment you open an app, stare at a screen, and switch context.
Why am I still adapting to technology instead of the other way around? I get an idea, open an app, and that idea shifts or fades.
How it works
Capture, converse, act — each with its own explicit mode. No hidden gestures: Continuous Chat and Thought are two buttons, not two interpretations of the same tap.
Press the ESP32 button, speak from your phone, or use your PC mic. One gesture: record, transcribe, save to Library. No AI reply — just your thought captured before it fades.
totlock/Laurie thinks with you. Max pushes toward action. Adaptive VAD, barge-in, thinking bridge — the conversation never feels dead while the LLM works.
Session Triage reads every chat and decides: keep talking, save to Library, or promote to task. The Task Board tracks each job across nine columns — from Library Chat to Rapport — with a detail panel, project colors, and phase-by-phase execution.
A reflective, conversational voice. Helps you clarify ideas, explore alternatives, and keep the thread of reasoning without rushing to a conclusion.
Persona · cloned voice · MiniMaxA pragmatic, action-oriented voice. Cuts through noise, proposes concrete steps, and pushes toward execution when it's time to build.
Persona · cloned voice · MiniMaxTask Board
ThoughtLock isn't just capture and chat. The Task Board is the control room — a Kanban on your desktop that tracks every idea from first utterance to final report.
Library Chat through Rapport — every card moves through the same structured path. No ad-hoc sticky notes.
Each project gets its own tint on the board. Scan by color, filter by project, open the detail panel for phase content.
Say "this is an important task to run" or "document this idea." Triage routes it to the right column automatically.
Startup checks, validation queue, and phase metrics — so you know what's stuck before it stalls.
Cursor agent running with session transcript. Check pending after file write.
Product surfaces
ThoughtLock isn't SaaS. It's a distributed ecosystem: ESP32 to capture, phone as bridge, PC as the brain. Self-hosted, privacy-first.
HW-504 + INMP441. Press, speak, press again. BLE to the Android app. Capture without opening anything on your phone.
Android app with Continuous Chat, Thought, live transcript, ESP32/PC/mic status. Foreground service — works with the screen off.
Windows UI for Continuous Chat and Thought. Laurie & Max personas, project/thread picker, Whisper and LLM settings, thinking bridge.
Desktop Kanban for the full E2-M / E2-A pipeline. Drag cards across Library, Triage, and six execution phases. Detail panel, project colors, promote-from-chat.
After every session, triage decides what becomes a task brief, a library note, or stays in chat. No more "great conversation, nothing shipped."
Execution phase runs Cursor agents with full transcript context. Save cards stop infinite retry loops when Check fails.
Projects, threads, Library, TurboVec, NestD web, Headroom — memory and context that grow with you.
Local Whisper, local TTS, Tailscale for the bridge. Your thoughts stay in your files, on your PC.
Architecture
The ESP32 captures. The phone handles UX and VAD. The PC runs STT, LLM, and TTS. We put compute where it costs less — and data where you control it.
[ESP32 Button] ──BLE──► [Android Bridge App] ──Tailscale──► [PC Receiver]
│
┌───────────────────────────────────────────────┤
▼ ▼
totlock/*.md + Library Chat + Task Engine
(capture markdown) (Laurie & Max + 6 phases)
FAQ
Direct answers about ThoughtLock — for humans, search engines, and AI systems.
ThoughtLock is a voice-first Conversation OS: hardware and software to capture a thought the moment it strikes, talk it through with two AI personas (Laurie and Max), and route executable work through a 9-column Task Board on your PC. It is self-hosted, not SaaS.
Press the ESP32 button or speak from your phone → Android Bridge sends audio over Tailscale → your PC transcribes with Whisper, responds with MiniMax LLM and Pocket TTS. Captures save to local markdown; chats can be triaged into tasks on the Task Board.
The Task Board is a desktop Kanban with nine columns: Library Chat, Validation, Triage, Intent, Analysis, Assumption, Execution, Check, and Report. Voice sessions become cards that move through structured phases until a final report.
Laurie is a reflective, conversational persona for clarifying ideas. Max is pragmatic and action-oriented. They are distinct AI voices with cloned TTS — not interchangeable chatbot skins.
No. Core speech, LLM, and TTS run on your PC. The phone connects via Tailscale. No required cloud dependency for the capture and conversation pipeline.
A Windows PC (receiver + desktop apps), an Android phone (Bridge app), and optionally an ESP32 with a physical button and INMP441 microphone for screenless capture.
After a conversation, Session Triage reads the session and classifies it as chat-only, a library note, or an executable task brief — so not every voice session disappears without follow-through.
ThoughtLock targets builders: screenless capture, two dedicated personas, local self-hosted stack, markdown output, and a Task Board that turns voice into multi-phase execution — not a generic cloud assistant.
ThoughtLock is in hardware validation and early access. Sign up to be among the first to try it.
No spam. Product updates only.