The Cue Complete field guide

A little setup.
A clearer talk.

Start with the notes you already have. Find the right format, build your checklist, and walk into rehearsal ready.

01 / Import formatting

Your outline, as a checklist.

Use a plain-text (.txt) or Markdown (.md) file saved as UTF-8. Put each talking point on its own line. For a predictable outline, use a dash and a space before each cue.

01

Write short cues

Use one idea per line. A cue is a reminder of what to cover.

02

Group the details

Indent a bullet with two spaces to add a sub-cue under the preceding cue.

03

Import and review

In the presentation editor, under Talking points, choose “Import Markdown or text file.”

talking-points-example.md
Download .md ↓
- Explain the problem
  - Share a customer example
  - Describe the impact
- Introduce the solution
- Close with the next step
Your imported checklist
  • Explain the problem
  • Share a customer example
  • Describe the impact
  • Introduce the solution
  • Close with the next step

Have a multi-part point? Indented bullets become sub-cues. The parent is set to require all its sub-cues. Review the imported outline before saving.

Which other formats can I use?

Plain lines and numbered lists also work. Bullets can start with -, *, +, or • followed by a space. A line wrapped in bold, such as **The problem**, creates a parent for the unordered bullets that follow. Use indented bullets for the clearest hierarchy; deeper indentation still creates direct sub-cues, not extra nesting levels.

Can I paste notes or import a PDF?

You can paste text under Talking points. Text-based PDFs can also be imported, but extracted line breaks may need cleanup. Scanned PDFs need conversion to selectable text first. For the most predictable results, use .txt or .md. Lines beginning with Presentation:, Target:, or Duration: are skipped; enter those details in the editor.

02 / Slide notes

A heading for every slide.

Slide notes use a separate import. Start each slide with a Markdown heading (# followed by a space). The text underneath becomes that slide’s speaker notes, up to the next heading.

slide-notes-example.md
Download .md ↓
# The problem
Share a customer example.
Describe the impact.

# The solution
Explain what changes and why.

# Next steps
Ask for a decision.
Your imported slides

The problem

Share a customer example.
Describe the impact.

The solution

Explain what changes and why.

Next steps

Ask for a decision.

Under Slide notes (optional) in the presentation editor, choose Import Markdown slide notes. Review the titles and speaker notes before saving.

Two imports, two jobs. Talking points build the checklist. Slide notes organize speaker notes; they do not add cues to your checklist or import a PowerPoint or Keynote deck.

03 / Your first session

Ready, when you are.

  1. Create a presentation, give it a title, and add or import your talking points.
  2. Review the outline. Edit, remove, or reorder cues and add optional slide notes, then save.
  3. Check the app’s speech and local-model readiness information. Use the manual checklist when automatic coverage is unavailable.
  4. Speak naturally. Pause when needed, and use manual check or uncheck controls to correct a cue.
  5. Review the session report and use the missed points to shape your next rehearsal.

Covered stays Covered. Once an automatic cue is covered, it stays covered for that session, including pause and resume. You can change its status manually.

04 / How points get checked off

Your meaning.
Your words.

A point gets an automatic checkmark when the recognized speech supports every completion requirement. You can use your own phrasing.

From speech to checkmark

  1. You define the point

    List what must be communicated in Completion requirements. If you leave that list empty, the point’s title and coverage detail define what is needed.

  2. The app reads recognized speech

    Apple’s on-device language model compares the meaning with each requirement, using recent speech and supporting passages retained for that unfinished point.

  3. Supporting words are checked

    The app checks that quoted evidence exists in the transcript. It also checks required numbers and directions of change, and looks for missing information or conflicts before adding a checkmark. Required causes get an additional review for contradictory timing or a denied effect.

What the states mean

Open

No relevant speech has been verified yet. A review may still be pending.

Emerging

Relevant speech was found, but completion is not verified. You may have named the topic, covered only part of it, or spoken words that need another check.

Covered

Every requirement has supporting speech and the completion checks passed. It stays Covered for this session, including pause, resume, and recovery, unless you change it manually.

Emerging is a progress signal. It is not a percentage or a checkmark. Unfinished points can change as recognition is revised.

An example in Balanced mode

Point: Explain the launch delay and our response.
Requirements: Explain that a staff shortage delayed the launch, and explain hiring contract staff as the response.

Emerging · topic only
“Let’s talk about the launch delay.”

The subject is named; the explanation is still missing.

Emerging · part covered
“We couldn’t launch on time because we were short-handed.”

The cause is explained; the response is still missing.

Covered · both requirements
“We couldn’t launch on time because we were short-handed. To catch up, we’re bringing in temporary help.”

Different wording communicates both required ideas.

Illustrative examples. Actual results depend on the recognized words and the local model’s assessment.

How close is close enough?

There is no fixed similarity percentage. A brief paraphrase can count if it communicates every required fact and relationship.

Balanced
Recognizes the required meaning, including paraphrases.
Strict
Asks for an explicit explanation.
Mention only
Asks the model to accept a clear mention for each requirement. Transcript evidence checks still apply.

Naming a topic, promising to explain it later, or asking whether a fact is true does not normally establish that fact. A question can count when the requirement itself is to ask or request something.

“Eight percent” can express “8%,” but “eighteen percent” cannot satisfy an 8% requirement. For a required cause, an unrelated event or an event that happened afterward does not explain the result.

If a point stays unfinished

Check the recognized words and the point’s review note where available. State any missing fact or connection clearly. A low speech-confidence warning blocks new automatic completion; a busy model can delay review. The model can also miss a valid explanation.

You can always check or uncheck a point manually. If the local model is unavailable, the checklist remains manual. Automatic coverage runs on-device without a cloud fallback. Read about on-device processing →

Write concrete requirements. “Explain that a staff shortage delayed the launch” gives the app more context than “Discuss the delay.” Groups follow their configured sub-cue completion rule.

05 / A few quick fixes

Get back to your talk.

My import is one long cue.

Put each idea on a separate line, preferably as a bullet. Visual word wrapping in your writing app is not the same as inserting a line break. Save a plain-text or Markdown file and import again.

My slide notes will not import.

Add a Markdown heading before each slide, such as # Introduction. Use “Import Markdown slide notes” under Slide notes, rather than the Talking points importer. Save the file as UTF-8 .md.

Why are there extra cues after importing again?

Imports append to the existing outline or slide list. Remove unwanted entries in the editor before saving, including duplicates from a second import.

My PDF contains no extractable text.

The file may be a scan or an image. Copy selectable text from the original document, or convert the scan to text, then save it as .txt or .md. Check the line breaks before importing.

Automatic coverage is unavailable.

Check the readiness information shown in the app and the recognition privacy settings. You can continue with the manual checklist. Read about on-device processing →