Art direction
Unifies lighting, styling, setting, subject, and mood into one visual intent.
Prompt phrase: Give the image strong art direction: deliberate styling, cohesive palette, and a clear emotional tone.
Complete vocabulary index
AI prompt vocabulary for people who know what they mean but not what to call it.
Browse 63 expert terms across image generation, resilient app builders, UX friction, customer discovery, and buyer messaging. Each term has a plain meaning and a prompt-ready phrase.
Camera angle
Controls power, intimacy, realism, and the viewer's relationship to the subject.
Prompt phrase: Use a slightly low eye-level camera angle to make the subject feel present without looking staged.
Cinematic lighting
Adds directional light, contrast, and intentional shadow like a film frame.
Prompt phrase: Use cinematic lighting with motivated key light, soft shadows, and controlled contrast.
Color grading
Gives the image a consistent mood through color and contrast choices.
Prompt phrase: Apply restrained color grading with natural skin tones, deep neutrals, and one quiet accent color.
Composition
Defines where the subject sits and how the eye moves through the image.
Prompt phrase: Use balanced composition with clear subject hierarchy and intentional negative space.
Depth of field
Controls what is sharp and what fades into the background.
Prompt phrase: Use shallow depth of field with the subject sharply focused and background softly separated.
Editorial photography
Signals a real-world, polished image that feels intentionally art directed.
Prompt phrase: Make it feel like editorial photography for a premium magazine, not generic stock imagery.
Lens choice
Gives the generator a familiar camera feel for intimacy, realism, or compression.
Prompt phrase: Use a 35mm documentary look or 50mm portrait lens depending on intimacy and realism.
Material texture
Makes objects feel physical instead of flat or synthetic.
Prompt phrase: Emphasize material texture, subtle surface imperfections, and realistic light interaction.
Negative prompt
Tells the generator what defects, styles, or objects to avoid.
Prompt phrase: Negative prompt: generic stock photo, distorted hands, plastic skin, clutter, text artifacts, oversaturated color.
Photorealism constraints
Constrains the generator away from common artificial-looking defects.
Prompt phrase: Constrain the image toward natural skin, realistic hands, believable lighting, and no synthetic polish.
Production design
Defines the setting, objects, clothing, and surfaces that make an image feel intentionally built.
Prompt phrase: Define the setting, props, wardrobe, and surfaces so the image feels art-directed, not generic.
Reference mood
Names the closest visual world without needing a copyrighted or exact reference.
Prompt phrase: Describe the visual reference as documentary editorial, premium magazine profile, or commercial campaign.
Shot type
Sets how much of the subject and scene the image should show.
Prompt phrase: Specify shot type: close-up, medium shot, wide shot, or establishing shot.
Checkpoint
Marks a safe place where users or background work can resume without replaying unsafe side effects.
Prompt phrase: Create checkpoints after durable steps so users can resume safely without replaying committed side effects.
Compensating action
Repairs or reverses side effects when a later step fails.
Prompt phrase: Identify compensating actions for each side effect if the workflow fails midway.
Draft persistence
Saves unfinished work safely so users can leave, refresh, or return without losing context.
Prompt phrase: Persist draft state after every meaningful change with versioning and safe resume behavior.
Durable workflow
Keeps long-running work recoverable across crashes, retries, and restarts.
Prompt phrase: Evaluate whether this should be a durable workflow with persisted steps and replay-safe handlers.
Error boundary
Contains UI failures so the rest of the app remains usable.
Prompt phrase: Place error boundaries around risky sections and define local recovery actions.
Fallback UI
Shows a useful interface when data, network, or a component fails instead of leaving users stranded.
Prompt phrase: Design fallback UI for loading, partial data, API failure, retrying, and manual recovery states.
Guard condition
Prevents a flow from moving forward unless the required facts are true.
Prompt phrase: Define guard conditions for each transition, including validation failures and disabled actions.
Idempotency
Makes repeated attempts produce the same result instead of duplicate side effects.
Prompt phrase: Add idempotency keys so retries and double submits cannot create duplicate side effects.
Invariant
Captures rules that must remain true across every state and transition.
Prompt phrase: List the invariants this flow must preserve, then test each transition against them.
Observability
Makes hidden failures visible through logs, metrics, traces, and events.
Prompt phrase: Add observability events for each transition, failure, retry, and recovery path.
Optimistic UI
Updates the interface before the server confirms, with a plan for failure correction.
Prompt phrase: Design optimistic UI states for pending, confirmed, failed, and rolled back changes.
Race condition
Catches bugs where timing or concurrent actions create the wrong final state.
Prompt phrase: Handle rapid clicks, two tabs, stale saves, and out-of-order responses without overwriting newer state.
Recovery path
Defines what happens after a failure so the user can continue instead of getting stranded.
Prompt phrase: For each failure mode, define what the user sees, what is retried, and how they can continue.
Resumability
Lets a user or job continue from a safe checkpoint instead of starting over.
Prompt phrase: Design resumability with saved checkpoints, versioned draft state, and clear restart rules.
Retry policy
Defines when to retry, when to stop, and how to avoid making failures worse.
Prompt phrase: Specify retry policy with timeout, backoff, maximum attempts, and non-retryable errors.
Source of truth
Defines which place owns the real state when UI, server, URL, and draft storage disagree.
Prompt phrase: Define one canonical source of truth for builder state across UI, server, URL, and saved draft.
State reconciliation
Resolves disagreements between UI state, server records, URL state, and saved drafts.
Prompt phrase: Add state reconciliation rules for UI state, server state, URL state, saved drafts, and stale writes.
Statechart
Models allowed states, transitions, guards, and side effects in one visible structure.
Prompt phrase: Model this as a statechart with explicit states, events, guards, and invalid transitions.
Step state
Names the lifecycle of each step so the app knows what is editable, complete, blocked, or recoverable.
Prompt phrase: Define step state for every screen: not started, active, valid, blocked, complete, failed, and recoverable.
Timeout budget
Defines how long each operation may wait before the product moves to retry, fallback, or recovery.
Prompt phrase: Set timeout budgets for each network operation, then define the retry, fallback, and recovery behavior.
Transaction boundary
Separates temporary edits from committed side effects so partial failure does not corrupt the workflow.
Prompt phrase: Separate temporary draft changes from committed side effects so partial failure does not corrupt progress.
Transition table
Makes every allowed and blocked movement between states explicit.
Prompt phrase: Create a transition table that lists every current state, event, next state, and rejected event.
Activation moment
Identifies the first point where users experience concrete value.
Prompt phrase: Identify the activation moment and reduce the steps required to reach first value.
Affordance
Describes the visual cue that tells users what action is possible.
Prompt phrase: Review the key affordances and make primary actions visually obvious without explanatory text.
Empty state
Turns a blank product moment into a useful next action.
Prompt phrase: Design empty states that show the next meaningful action and reflect the user's current context.
Error taxonomy
Groups failures so the product can respond with the right message and recovery path.
Prompt phrase: Create an error taxonomy with user-fixable, system-retryable, and support-required failures.
Information architecture
Clarifies how information, screens, and actions are grouped.
Prompt phrase: Audit the information architecture and propose a clearer grouping of screens, objects, and actions.
Jobs to be done
Frames the user's goal as progress they are trying to make, not just a feature request.
Prompt phrase: Rewrite this as jobs to be done with situation, motivation, desired progress, and current alternatives.
Progressive disclosure
Shows complexity only when it becomes relevant.
Prompt phrase: Use progressive disclosure so advanced choices appear only after the user has enough context.
User mental model
Names the user's internal picture of how the product should work.
Prompt phrase: Map the user's mental model and compare it against the product's current object model.
Message hierarchy
Orders claims so the most important point lands first.
Prompt phrase: Create a message hierarchy from primary claim to proof points to objections.
Objection handling
Answers the reasons a user might hesitate.
Prompt phrase: List likely objections and write concise responses backed by proof, not hype.
Positioning
Places the product in a buyer's mind against alternatives.
Prompt phrase: Write positioning that names the category, target customer, core pain, alternative, and sharp difference.
Tone of voice
Controls how the writing feels while keeping the message intact.
Prompt phrase: Rewrite with a clear tone of voice: direct, specific, calm, and credible.
Value proposition
States the outcome and why it matters to a specific person.
Prompt phrase: Turn this into a value proposition with audience, pain, promised outcome, and reason to believe.
Assumption mapping
Sorts beliefs by importance and uncertainty so the next validation step is not arbitrary.
Prompt phrase: Create an assumption map that ranks beliefs by importance, uncertainty, evidence, and next test.
Concierge MVP
Delivers the value manually first so you can learn before investing in automation.
Prompt phrase: Validate with a concierge MVP: deliver the outcome manually, capture objections, and only automate repeated demand.
Customer segment
Narrows research to a group with shared context, pain, budget, and behavior.
Prompt phrase: Define customer segments by shared pain, current workaround, urgency, and buying context.
Demand signal
Captures observable behavior that suggests people may actually want the product, not just say it sounds interesting.
Prompt phrase: Separate weak interest from demand signals: repeated pain, active workaround, budget owner, urgency, and willingness to take a next step.
Evidence standard
Defines how much evidence is enough to act.
Prompt phrase: Define the evidence standard required before making this product decision.
Fake door test
Measures real interest in an offer before building the full product.
Prompt phrase: Use a fake door test to measure demand before building: offer the feature, track intent, and disclose honestly when needed.
Hypothesis
States what you believe and what evidence would change your mind.
Prompt phrase: State the hypothesis, why it matters, what evidence supports it, and what would disprove it.
Interview guide
Structures conversations around real behavior instead of leading questions.
Prompt phrase: Create an interview guide with behavior-first questions and no leading prompts.
Problem-solution fit
Tests whether a real customer segment has a painful problem and believes the proposed solution is worth trying.
Prompt phrase: Evaluate problem-solution fit by separating customer pain, current alternatives, proposed solution, and evidence quality.
Research question
Turns vague curiosity into a testable learning target.
Prompt phrase: Convert this into research questions that can be answered through interviews, desk research, or usage data.
Riskiest assumption
Names the belief that would kill the idea fastest if it turned out to be false.
Prompt phrase: Identify the riskiest assumption first, then design the smallest test that can disprove it.
Stage-gate process
Breaks idea evaluation into explicit gates with entry criteria, evidence requirements, and continue/stop decisions.
Prompt phrase: Design this as a stage-gate process with entry criteria, evidence required, decision outputs, and stop/continue rules for each gate.
Switching cost
Captures the friction that keeps people on their current solution.
Prompt phrase: Analyze switching costs, including setup effort, data migration, team habits, trust, and risk.
Willingness to pay
Tests whether the pain is strong enough that a real buyer would spend money, time, or political capital.
Prompt phrase: Test willingness to pay by asking about current spend, budget owner, urgency, alternatives, and the cost of doing nothing.