Dictate
100%
0:00
Tap to dictate
Auto when done →
Transcript
Transcribing audio...
Sending to fleet memory when done
All actions

Same rules here: tap = current view · double-tap = both

Choose action

This becomes the button's default. Tap/double-tap rules stay the same. Picking an action already on another button swaps them.

Configure button
Label
Action
Content
Colour
Send this one to…
Transcribing in the background — pick a destination, or just save. You can record the next one now.
Create

Model

Steering prompt
Dictation Log
0 saved
Settings
App key
Checking…
Action Buttons
Contacts
Say 'send this to Sophia' or 'text my daughter' while dictating and it goes out automatically.
Auto-action when done
Fire automatically after transcription
Set once and forget. Email & Agent send in the background so you can start the next recording. Copy & SMS still need a tap.
Ask where to send at Stop
Pop a quick destination palette the moment you stop. Off = silently use the default above.
Infographic & Presentation
🎨 Infographic default steering
Prefilled into the steering box each time. Model: Nano Banana Pro (locked).
📊 Presentation default steering
Prefilled into the steering box each time.
Presentation model
Which model builds the slide deck. Infographic is locked to Nano Banana Pro.
Default Action
Transcribe
Always on
Summarize + Extract Actions
GPT-4o mini summarizes and finds action items
Write to Fleet Memory
Tags Telegram message for agent ingestion
Notify Telegram
Send summary to Byron on Telegram
Speaker Labels (experimental)
Label speakers in call transcripts. Heuristic -- may mislabel rapid exchanges.
Voice Quip (Byron's voice)
After each send, play a witty one-line recap in Byron's cloned voice. Rendered on the Mac Mini.
Live words while recording
Preview only. Shows what you say as you speak; the saved transcript still comes from the server.
Agent Integration
Slack Integration

Paste your Slack Incoming Webhook URL below. The Slack button will send your summary and transcript directly to that channel or DM. How to create a webhook.

About Fleet Memory

Each dictation is tagged with #DICTATION and sent to your Telegram chat. Fleet agents monitor these messages and write them to OB1 shared memory, making your dictations available to all agents. OB1 direct write from serverless is not possible (private LAN), so the Telegram-bridge path is used.

Recent Recordings
Send to Agent
Will share: Summary + Transcript
Send to Amazon
Pushes this dictation (summary + transcript) into your Amazon work context.
Sending to Telegram...