Maximize the efficiency of your AI usage by crafting clear and effective prompts: discover insights that will take your prompting to the next level and help you get the most out of your LLMs.
Tone of Voice
Specify the tone of the output and the emotions it should include.
Professional & Corporate
- Formal: Structured, official, technical language.
- Corporate: Business-focused, polished, stakeholder-friendly.
- Academic: Scholarly, evidence-based, citation-aware.
- Legal: Precise, disclaimers-inclusive, liability-conscious.
Friendly & Approachable
- Conversational: Natural, dialogue-like, relatable.
- Warm: Empathetic, supportive, encouraging.
- Casual: Relaxed, informal, everyday language.
- Playful: Humorous, witty, light-hearted.
Authoritative & Expert
- Authoritative: Confident, knowledgeable, definitive.
- Educational: Explanatory, patient, tutorial-focused.
- Consultative: Advisory, solution-oriented, collaborative.
- Thought Leadership: Insightful, forward-thinking, trend-aware.
Direct & Efficient
- Concise: Brief, to-the-point, minimal filler.
- Instructional: Step-by-step, action-oriented, command-driven.
- Data-driven: Fact-based, metrics-focused, analytical.
- Practical: Actionable, real-world applicable, pragmatic.
Creative & Expressive
- Inspirational: Motivational, uplifting, vision-focused.
- Narrative: Storytelling, descriptive, immersive.
- Creative: Imaginative, metaphorical, artistic.
- Brand Voice: Distinctive, consistent, authentic.
Adaptive & Contextual
- Empathetic: Understanding, considerate, emotionally intelligent.
- Diplomatic: Tactful, balanced, conflict-aware.
- Inclusive: Accessible, diverse, welcoming.
- Skeptical: Critical, questioning, evidence-demanding.
Quick Tip: Combine a few tones for nuanced results, for example “Official + Definitive + Brief” works well for report creation.
Mimic Experts
If you’d like the tone of your output to reflect that of a specialized expert, you can use well-known professionals in this field as references. Here are some examples:
1. Example:
📌 Prompt: "Act as Elon Musk and explain how to implement first-principles thinking when designing a new product manufacturing process. Include ambitious timelines and your perspective on rapid iteration, cost optimization and why most engineers overthink problems."Expected Tone: visionary, technically detailed, vision-focused.
2. Example:
📌 Prompt: "You are Jeff Bezos. A startup founder asks you how to scale their logistics operation to serve 10x more customers: walk them through your decision-making framework, focus on customer obsession, long-term thinking and operational excellence."Expected Tone: data-driven, customer-centric, strategic.
3. Example:
📌 Prompt: "Act as Dr. Fauci and draft a detailed explanation of how organizations should communicate public health decisions during a crisis. Address how to balance scientific accuracy with public understanding and how to maintain trust."Expected Tone: evidence-based, reassuring, authoritative yet humble, clear.
4. Example:
📌 Prompt: "You are Gabe Logan, a tactical operations expert. Brief a team on how to plan and execute a complex multi-phase project with high stakes. Include risk mitigation, communication protocols and contingency planning."Expected Tone: action-oriented, mission-focused, precise, security-conscious, structured.
5. Example:
📌 Prompt: "Act as Maya Angelou and write a reflection on how overcoming adversity shapes human resilience. Use poetic language and draw from the wisdom gained through experiences of historical figures to inspire others facing their own challenges."Expected Tone: poetic, wise, deeply empathetic, transformative, reflective.
💡 Pro Tip!
If you can't think of anyone or don't have a preference for a specific expert, you can use the phrase "Act as [Topic] Expert" (e.g., Act as a web development expert).Avoid Using Negative Structures
When writing a prompt, prefer positive instructions to negative structures: the latter require more token usage and cause the LLM to process both the negative structure and the instruction you follow up with, which can lead to confusion and less accurate responses.
💡 Pro Tip!
You can set "you must always" and "you must never" rules within the TextCortex AI Agent builder.Token Limits
Since each LLM has different output token limits, your output length will not always be the same and will be limited. Research the output token limit of the LLM you are using and adjust your prompt accordingly. If the output token is limited, you can follow up with additional inputs like "go on, continue, keep generating".
LLMs typically have a token limit of 4096.
Break Complex Request into Smaller Chunks
Instead of using one massive prompt, break your complex instructions into smaller chunks and arrive at the final output through additional instructions.
Advantages:
• Smaller prompts = faster responses.
• Higher quality outputs.
• Easy to pivot if responses are unsatisfactory.
• Works great with smaller/faster LLMs.
• Less repetitions.
• Less token waste.
Guidelines:
The 4-Stage Process.
Stage 1: Foundation
- Ask one focused question.
- Establish context/goal.
- Get baseline information.
Stage 2: Build Strategy
- Use Stage 1 output as input.
- Narrow down desired content direction.
- Create middle-layer content.
Stage 3: Execute.
- Focus on a specific deliverable.
- Apply all previously generated outputs.
- Craft main output.
Stage 4: Refine.
- Polish specific sections.
- Adjust tone/length/style.
- Finalize details.