Audience Adaptation

This dimension controls the specificity of language. Instructions should specify whether to use domain-specific terminology or provide plain-language explanations.

• For experts: Use precise technical terms (e.g., 'stochastic gradient descent', 'attention weights').
• For novices: Define acronyms on first use and use analogies (e.g., 'The model's focus mechanism, called attention...').
• Example Instruction: 'Explain quantum computing concepts using metaphors suitable for a high school student, avoiding mathematical notation.'

This dictates the granularity of detail. It involves instructing the model on the appropriate level of abstraction, from high-level overviews to step-by-step mechanics.

• Strategic Depth: For executives, focus on business impact, ROI, and high-level architecture.
• Tactical Depth: For engineers, detail APIs, algorithms, data flow, and error handling.
• Example Instruction: 'Provide a summary of the solution's benefits in three bullet points. Then, in a separate section labeled "Technical Deep Dive," explain the consensus mechanism used.'

The choice of in-context examples or hypothetical scenarios must resonate with the audience's domain. This grounds abstract concepts in familiar contexts.

• For a finance audience: Use examples about risk modeling, portfolio optimization, or fraud detection.
• For a healthcare audience: Use examples about patient diagnosis, clinical trial analysis, or HIPAA compliance.
• Example Instruction: 'When illustrating the concept of an API, use an analogy related to restaurant ordering (customer, waiter, kitchen) for a general audience, or a direct cURL command example for developers.'

This is an explicit declaration within the prompt of what foundational concepts the model can assume the user understands. It prevents over-explaining basics or under-explaining prerequisites.

• Explicit Boundary: 'Assume the user has a computer science degree and understands basic data structures.'
• Implicit Calibration: The model infers knowledge from user queries but can be guided: 'If the user asks about transformers, do not explain basic neural networks unless they explicitly ask.'
• **This acts as a form of knowledge boundary, scoping the conversation's starting point.

While often managed by a tone modulator, audience adaptation specifically links formality to professional context and perceived authority.

• For C-Suite: Concise, confident, decision-focused. Avoid casual language.
• For peer developers: Collaborative, can include technical shorthand and accepted informalisms.
• For customer support: Empathetic, patient, and clear.
• Example Instruction: 'Adopt the tone of a senior consultant delivering a board-level briefing: authoritative, data-driven, and focused on strategic implications.'

This dimension adapts the response schema and organization to the audience's consumption preferences.

• For quick scanning: Use bold headings, bullet points, and a TL;DR summary.
• For detailed analysis: Provide structured reports with sections, tables, and appendices.
• For integration: Enforce a strict JSON schema for developer consumption.
• Example Instruction: 'Format the response in two parts: 1) an "Executive Summary" in three sentences, and 2) a "Technical Specification" using Markdown headers and code blocks.'

The specification of a persona or functional identity (e.g., 'senior software architect', 'patient educator') within a system prompt. This is the foundational step for audience adaptation, as the defined role inherently carries assumptions about expertise, goals, and communication style that the model should emulate.

• Example: You are a cybersecurity expert explaining concepts to a board of directors. This single directive implicitly triggers adaptation by setting the role's knowledge depth and the need for business-relevant analogies.

A directive within a system prompt that explicitly sets the desired communication style. While audience adaptation defines the who, tone modulation defines the how of the interaction. These are often used in conjunction.

• Key Tones: Formal, casual, empathetic, concise, persuasive, neutral.
• Example for Adaptation: Explain this medical concept with a reassuring and empathetic tone suitable for a newly diagnosed patient.

The process of defining and limiting the set of tasks a model is instructed to perform. Effective audience adaptation requires precise scoping to match the user's expected needs and prevent the model from over-explaining or venturing into irrelevant domains.

• Mechanism: Explicitly listing in-scope and out-of-scope topics based on the audience.
• Example for a Novice Audience: Your role is to answer basic 'how-to' questions about Python syntax. Do not provide deep dives into algorithm optimization or systems design.

An explicit instruction that defines the scope or limits of information a model should use. This is critical for audience adaptation to ensure explanations are appropriately grounded and do not reference concepts beyond the user's presumed level.

• Common Forms: Only use information provided in the context below. or Assume the user has no prior knowledge of calculus.
• Prevents: The model from using advanced jargon or concepts without first defining them.

A constraint placed in a system prompt that instructs the model to limit the length of its response. This is a direct, quantitative lever for adaptation, forcing conciseness for expert users or allowing elaboration for beginners.

• Implementation: Provide your answer in under 100 words. or Give a comprehensive explanation using up to 500 tokens.
• Engineering Consideration: Must be balanced with the need for complete and clear explanations.

A prompt directive that uses if-then logic to dictate different model behaviors based on specific characteristics of the user input. This enables dynamic, runtime adaptation within a single system prompt.

• Structure: If the user asks a follow-up question indicating confusion, provide a simpler analogy. If they ask for technical specifics, provide detailed parameters.
• Advanced Use: Can be combined with metadata (e.g., user role from an API) to trigger different adaptation paths automatically.