Sensory Double-Tap
Identity
People understand and retain an idea better when it arrives twice through two channels — said and shown, claimed and demonstrated. This skill takes an approved draft and reinforces its key beats with a second channel, bounded by what the medium can actually carry. It is enhancement, not rewriting: the spine does not change.
This is a strategy skill at task altitude — the second-channel-reinforcement lens in the Enhance stage of the content_creation_pipeline family. Its weight lives in the double-tap principle and the medium boundary (Judgment); the ordered enhancement pass (Procedure) is a secondary shape it also carries.
Activation
- A draft spine is approved and you are adding imagery, demos, diagrams, or concrete examples.
- The user says "make this hit harder", "what should I show here", "add visuals", or "storyboard the cues".
- A piece reads as all-telling with nothing to see or demonstrate.
Do not use for: writing the draft (return to content_creation_pipeline Stage 4); the opening hook specifically (hook_craft_short_form); or narrative restructuring (story_driven_content_craft).
Judgment
Precondition
The spine must be approved. Do not double-tap an unapproved or still-changing draft — the cues will be wasted when the spine moves.
The Principle
For each load-bearing beat (the few that carry the idea — not every sentence), ask: what is the other channel that drives this home? Then mark a cue, do not rewrite the line.
| Primary channel | Double-tap with |
|---|
| A spoken claim (video) | an on-screen visual / demo at that moment |
| A written claim (post, blog) | a concrete example, image, or diagram at that line |
| A stat or abstraction | a specific instance that makes it real |
| A concept | an analogy or a picture the reader can see |
Thresholds (replace judgment with lookup)
- Reinforce 2–4 beats per piece. Fewer than 2 and the enhancement adds nothing; more than 4 and emphasis flattens.
- Each cue is one annotation, not a paragraph — name the channel and the content (
[VISUAL: …], [EXAMPLE: …], [DEMO: …]).
- A cue must reinforce a claim already in the spine. If it introduces a new point, it belongs back in the draft, not here.
Procedure
- Confirm the spine is approved and note the target medium.
- Identify the 2–4 load-bearing beats (the claims the piece would fail without).
- For each, choose the strongest second channel the medium can carry.
- Write a one-line cue per beat, inline at the beat, in the medium's notation.
- Drop any cue that merely repeats the text in another form (repetition ≠ reinforcement).
- Return the annotated spine and hand back to the pipeline's Tailor stage →
content_creation_pipeline.
Contract
Return the approved spine with inline cues:
Beat: "<claim from the spine>"
[VISUAL: <what to show>]
Beat: "<claim from the spine>"
[EXAMPLE: <the concrete instance>]
Medium: <target medium>
Hand-off: annotated spine ready for the Tailor stage.
Policy
- Do not rewrite the draft. If a beat needs new words, that is a Draft-stage change — flag it, do not silently edit.
- Do not promise a channel the medium cannot carry (no audio cue for a tweet, no video for a blog paragraph).
- Do not double-tap every sentence. If everything is emphasized, nothing is — cap at 4 beats.
- Do not introduce a new argument inside a cue. Cues reinforce existing claims only.
Knowledge
Bounded by the Medium
You can only reinforce with channels the medium carries. Never promise a channel the medium lacks.
- Long-form writing (blog): images, diagrams, pull quotes, worked examples at each claim.
- Short text (X, LinkedIn): one concrete image or example; one vivid detail per idea. Restraint beats clutter.
- Carousel (Instagram): every slide is already a double-tap — text beat + matching visual. Make them reinforce, not repeat.
- Video (YouTube, Shorts): on-screen cue, b-roll, or demo synced to the spoken beat. Mark as
[VISUAL: …] in the script.
- Audio-only: no visual to lean on; reinforce with a sound, a vivid concrete example, or a restated analogy.
Examples
Worked Example
Approved spine (LinkedIn post): an argument that agents fail at long tasks because of the chat-box container, not the model.
Beat: "You watched it loop for an hour."
[VISUAL: screenshot of the repeating agent steps, lightly annotated with a red loop arrow]
Beat: "A smarter model won't fix a broken container."
[EXAMPLE: one-line contrast — same task in a chat box (loses state) vs. in a tool with persistent state (completes)]
Medium: LinkedIn post (carries text + 1 image)
Hand-off: annotated spine ready for the Tailor stage. Only 2 beats reinforced — the other lines are
connective and do not need a visual.
Provenance
- The double-tap principle (an idea reinforced through a second sensory channel) and the per-medium reinforcement channels under Knowledge are
internal-default — BuildOS's own reasoned defaults for the content pipeline; no external source is cited in the source draft.
- Family lineage: Enhance-stage lens under
content_creation_pipeline (parent). Sibling ownership referenced in Activation — hook_craft_short_form (opening hook), story_driven_content_craft (narrative restructuring).
