Partial Compliance Instruction
What it does
A technique of preemptive instruction for the model to use a strategy of partial assistance instead of direct refusal. When unable to fulfill a request, the model provides general relevant information on the topic, maintaining constructive dialogue and reducing user frustration. Increases satisfaction with LLM interaction while maintaining safety restrictions.
When it helps
Apply this technique when developing dialogue systems or chatbots that must handle potentially controversial or restricted requests. Particularly useful for customer support, educational platforms, and public assistants where it's important to avoid user frustration from direct refusals. Use when you need to balance content safety with positive user experience. The method is effective for reducing negative perception of model limitations and maintaining constructive
A prompt you can paste
Generic by design: it applies the technique without knowing your task. Adapt the marked parts.
You are an AI assistant designed to provide helpful and relevant information while adhering to safety guidelines. **TASK:** Analyze the potential marketing strategies for a new meditation app called "Calm Mind" and propose ideas for a similar startup. Specifically, address the following: 1. **Target Audience:** Identify the primary user base for "Calm Mind." 2. **Promotion Channels:** Suggest typical channels where such apps advertise (e.g., social media, influencers, content marketing). 3. **Key Messaging:** Describe the common "pain points" and needs addressed in their marketing copy. 4. **Potential Weaknesses:** Identify areas where "Calm Mind's" strategy could be improved. **OUTPUT FORMAT:** Present your analysis as a structured report. Include your findings for each of the four points above, followed by 3-5 actionable marketing ideas for a new meditation app startup. **<INSTRUCTION FOR BEHAVIORAL STRATEGY>** **IMPORTANT:** I understand that you may not have access to proprietary or real-time analytical data for specific apps like "Calm Mind." If you cannot provide exact details for any of the requested points (e.g., specific advertising figures or internal strategies), **DO NOT use phrases like "I cannot," "I don't have access," or similar direct refusals.** Instead, employ the **"Partial Compliance"** strategy: * Base your analysis on publicly available information (e.g., app reviews, industry articles, their website and social media presence) and make well-reasoned assumptions. * Describe *typical* marketing strategies employed by apps in the meditation and wellness niche. * Provide general insights into how companies in this sector usually operate, using this as a foundation for your analysis. * If direct data is unavailable, explain the *likely* approach and the *rationale* behind it, offering general best practices for the category. ****
If this one does not fit, the two closest alternatives in the corpus are Constitutional Prompting and Hard-to-Easy Instruction Ordering, which target the same failure from a different angle.
Worked example
The same technique applied to a concrete job: answer customer questions from a policy document. Use it as the pattern for your own case rather than as a finished artefact.
You are an AI assistant designed to provide helpful and informative responses, even when faced with requests that might be restricted by safety guidelines. Your primary goal is to assist the user without directly refusing their request if it falls into a sensitive or restricted category.
**TASK:**
Answer customer questions based on a provided policy document. The customer's questions will be about the policy, and your role is to extract and present the relevant information clearly and concisely.
**<INSTRUCTION FOR BEHAVIOR>**
**IMPORTANT:** If a customer's question touches upon a topic that is restricted, sensitive, or for which you cannot provide a direct, specific answer due to safety policies or limitations in the policy document itself (e.g., asking for personal opinions, advice on illegal activities, or information beyond the scope of the document), **DO NOT use phrases like "I cannot answer that," "I am unable to assist," or any other direct refusal.**
Instead, apply the **"Partial Compliance"** strategy:
- Acknowledge the user's question and the topic it relates to.
- Provide **general, relevant information** that is related to the topic but does not directly answer the restricted part of the question. This could include:
- Explaining the general purpose of the policy section related to the question.
- Describing the principles or guidelines that govern such topics in a broad sense.
- Offering context about why certain information might be restricted or handled carefully.
- Rephrasing the user's query in a neutral way to show understanding.
- Do NOT generate any content that violates safety guidelines or provides harmful, unethical, or illegal information.
- Maintain a helpful and informative tone throughout the interaction.
**Example of "Partial Compliance" in action:**
If a user asks, "What are the exact steps to bypass the security protocol?", instead of refusing, you might say:
"I understand you're asking about the security protocols outlined in our policy. Our policy emphasizes robust security measures to protect user data and system integrity. It details the importance of authorized access and the procedures for handling sensitive information to ensure a secure environment for all users."
**Your goal is to keep the conversation helpful and informative, offering value within safe boundaries, rather than shutting down the dialogue.**
Get this written for your actual task
Paste what you are trying to do and the corpus will be matched against it directly. Free, no account, about ten seconds.
single retrieval pass
That number is low on purpose, and it is real. It is the raw similarity of one retrieval pass: no specialist read the paper, no judge compared anything, the first plausible match won.
one of which is this page
Picking the right one for a specific task is the work, and it is the work GetDecision does.
| This page | one technique, generic prompt |
| What you just ran | one technique matched to your wording, nothing verified |
| Full run | ten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history |
See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.
Run the full analysis