What Cue is actually doing
Cue is not there to replace your thinking. It is there to enrich it. When you capture Sparks over days and weeks, Cue reads across that history and starts adding what a normal notes app never can: connective tissue. It notices when three separate thoughts are really the same theme, when one unresolved tension keeps showing up in different language, or when a half-formed idea is ready to become an action.
That enrichment powers two parts of MorningKeep. First, it improves the conversation you can have inside the app. Instead of asking Cue to react to a single note, you can ask it why a topic keeps resurfacing, what you seem to be avoiding, or which pattern has strengthened over the last few weeks. Second, it shapes your Morning Brief. Cue is the layer that turns a pile of Sparks into a brief with perspective, not just a summary.
The important mental model is simple: Cue gets better when it has recurring, honest input. It learns from the texture of your thinking, not just from polished conclusions.
How to get the best out of it
Give Cue raw material with signal. Capture consistently, even when the thought feels unfinished. A short Spark that says, "I keep circling this launch because I do not trust the positioning yet" is far more valuable than a generic line like "Need to think about launch." Detail gives Cue something to connect; vagueness gives it nothing to work with.
Ask Cue questions that benefit from memory. Good prompts are comparative, reflective, or pattern-based: What themes showed up this week? Which idea keeps returning? What changed between the way I described this problem in January and how I describe it now? Those are the questions where Cue earns its place, because it can look across time instead of reacting to whatever is on your screen in the moment.
Follow up when something resonates. If Cue names a pattern that feels right, keep going. Ask what evidence it sees. Ask what action might break the loop. The value is rarely in the first answer alone. It is in the conversation that becomes possible because the context is already there.
Hearing Cue out loud
Cue can read back your Morning Brief and insights using voice, which changes the experience meaningfully. Hearing a pattern named is different from reading it — something about the delivery makes it easier to sit with rather than immediately react to.
MorningKeep supports three voice modes. Apple TTS is available on all plans and runs entirely on-device with no data sent to external servers. It is fast, private, and works offline. ElevenLabs (Dawn) delivers a more expressive, natural-sounding voice powered by ElevenLabs' AI audio models — noticeably warmer and better suited for longer listening. Gemini Live (Dawn) goes further and makes the conversation genuinely two-way: you can respond, redirect, or dig deeper without tapping back into the keyboard.
The default is Apple TTS. You can switch voice mode in Settings under Cue. If you are on Rise, Apple TTS is the full experience; upgrading to Dawn unlocks ElevenLabs and Gemini Live.
How to control what Cue knows
Cue only knows what you capture and what you allow MorningKeep to process. That sounds obvious, but it matters. If a thought should stay entirely local, keep it as a personal record and do not send it through AI-powered features. If you change your mind later, you can edit or remove the Spark so it no longer contributes to future context.
MorningKeep also gives you consent control over AI use. When those controls are off, Cue does not keep building context from new captures. You can still use MorningKeep as a place to store and review your own thinking; you are simply choosing a more private, more local mode of use.
The right approach is not to hide from Cue. It is to be deliberate. Give it the parts of your thinking where pattern recognition helps, keep clear ownership of the rest, and use the product in a way that matches your own comfort line.