Blog

Notes on voice-first work

Short, practical writing on dictation, Mac workflows, and getting thoughts into text.

Changelog
Local notes and cloud models can work together
BlogOct 4, 2026

Local notes and cloud models can work together

Learn what stays local, what crosses the network, and how to use selected note context with a cloud AI model without uploading your whole working record.

Mrigesh ParasharRead
How to speak an AI prompt that is easy to follow
BlogOct 4, 2026

How to speak an AI prompt that is easy to follow

Use a simple four-part order for voice prompts: goal, context, constraints, and output. Then add exact inputs and review before sending.

Mrigesh ParasharRead
What should an AI agent confirm before it acts?
BlogOct 4, 2026

What should an AI agent confirm before it acts?

A practical five-part check for AI agent actions: action, target, payload, consequence, permission scope, and a clear receipt after execution.

Mrigesh ParasharRead
Why screen context makes voice more useful
BlogOct 4, 2026

Why screen context makes voice more useful

Screen context helps voice tools understand what “this” refers to. Learn why the active app, selection, scope, and review matter.

Mrigesh ParasharRead
Your notes should be files you can keep
BlogOct 4, 2026

Your notes should be files you can keep

Learn what file ownership changes for your notes, what Markdown does not guarantee, and how to test whether a note remains useful outside its app.

Mrigesh ParasharRead
How to keep AI notes tied to their sources
BlogOct 3, 2026

How to keep AI notes tied to their sources

Keep AI-generated notes connected to recordings, transcripts, meetings, and documents with a simple source, transformation, claim, and review workflow.

Mrigesh ParasharRead
How to organize notes so AI can use them later
BlogOct 3, 2026

How to organize notes so AI can use them later

Organize project notes with clear titles, current state, decisions, sources, dates, and links—without building a high-maintenance second brain.

Mrigesh ParasharRead
Why Karpathy’s LLM Wiki makes sense for AI agents
BlogOct 3, 2026

Why Karpathy’s LLM Wiki makes sense for AI agents

A plain explanation of Karpathy’s LLM Wiki, how agents can navigate its knowledge, and why sources, updates, and human review still matter.

Mrigesh ParasharRead
Why Markdown is useful memory for AI work
BlogOct 3, 2026

Why Markdown is useful memory for AI work

Learn why Markdown works well as durable AI context, what it does not solve, and how to keep one project note useful across sessions.

Mrigesh ParasharRead
Rent your intelligence. Own your context.
BlogSep 30, 2026

Rent your intelligence. Own your context.

Your AI tools may change. Your knowledge should stay yours. Why context ownership shapes Cue’s local-first architecture, from voice capture to connected agents.

Mrigesh ParasharRead
Meeting Notes Without a Bot Joining the Call on Mac
BlogSep 29, 2026

Meeting Notes Without a Bot Joining the Call on Mac

Learn how bot-free meeting notes work on a Mac, how to check both sides of the audio, and what to verify before relying on a transcript.

Mrigesh ParasharRead
A Meeting Summary Template for Decisions, Actions and Open Questions
BlogSep 29, 2026

A Meeting Summary Template for Decisions, Actions and Open Questions

Copy a practical meeting summary template, see a worked example, and learn how to keep decisions, action items and unresolved questions separate.

Mrigesh ParasharRead
Why Cue gives you both on-device and cloud AI
BlogSep 9, 2026

Why Cue gives you both on-device and cloud AI

Cue gives you a fast, offline-capable local path and connected AI when a task needs more capability, while your durable work stays in files you control.

Mrigesh ParasharRead
How to turn a spoken brain dump into a useful note
BlogSep 9, 2026

How to turn a spoken brain dump into a useful note

Capture a messy voice thought without interrupting yourself, then use Cue Edit Mode to structure, rewrite, and review the useful part.

Mrigesh ParasharRead
AI Meeting Notes on Mac: From Conversation to Decisions and Action Items
BlogSep 7, 2026

AI Meeting Notes on Mac: From Conversation to Decisions and Action Items

Learn how to capture a meeting on Mac and turn the transcript into reviewed notes, decisions, action items, and clear next steps.

Mrigesh ParasharRead
How to turn your notes into working memory for Claude, Cursor, and Codex
BlogSep 7, 2026

How to turn your notes into working memory for Claude, Cursor, and Codex

A practical workflow for turning a few current notes into selective, source-linked working memory that an AI coding agent can actually use.

Mrigesh ParasharRead
Cue: A private, voice-first AI workspace for Mac
BlogSep 7, 2026

Cue: A private, voice-first AI workspace for Mac

Meet Cue: free on-device dictation, meeting capture, local Markdown notes, sourced AI, and voice editing in one private, voice-first Mac workspace.

Mrigesh ParasharRead
Voice input in Claude Code and Cursor on Mac: setup and Cue workflow
BlogAug 24, 2026

Voice input in Claude Code and Cursor on Mac: setup and Cue workflow

Set up voice input in Claude Code and Cursor on Mac, compare built-in dictation with Cue, and learn which prompt details to speak, type, and review.

Mrigesh ParasharRead
Voice input should work like a keyboard
BlogAug 20, 2026

Voice input should work like a keyboard

Voice should feel like dependable input on a Mac, with local processing available and cloud models used by choice.

Mrigesh ParasharRead
Free voice-to-text for Mac
BlogAug 12, 2026

Free voice-to-text for Mac

A short guide to when built-in dictation is enough, and when a dedicated Mac voice-to-text app makes more sense.

Mrigesh ParasharRead
How We Built Cue's Real-Time Transcription Pipeline
BlogMay 15, 2025

How We Built Cue's Real-Time Transcription Pipeline

A look under the hood at how Cue delivers sub-200ms voice-to-text with context-aware corrections.

Mrigesh ParasharRead
Why Voice Typing Is the Biggest Productivity Unlock You're Ignoring
BlogFeb 20, 2025

Why Voice Typing Is the Biggest Productivity Unlock You're Ignoring

You speak at 150 WPM. You type at 40. The math is obvious — so why aren't more people voice-typing?

Mrigesh ParasharRead