Documentation Index
Fetch the complete documentation index at: https://neuroai-cf1c4abf.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
The Golden Rule
Neuro AI is powerful, but it needs clear instructions. A well-structured prompt dramatically improves output quality, relevance, and usefulness.
Part 1: Fundamental Prompting Principles
Rule 1: Always Use the Three-Part Structure
Every effective prompt has three essential components:
[ROLE/CONTEXT] + [TASK] + [FORMAT] = Excellent Results
Example: Poor vs Good
❌ POOR:
"Analyze this spreadsheet"
"You are a financial analyst specializing in quarterly
reporting. Analyze the Q4 2024 sales spreadsheet attached.
Focus on: revenue trends by product line, variance analysis
vs Q3, and key growth drivers.
Present results as:
- 1-page executive summary
- Detailed findings with charts
- Top 3 recommendations with expected impact"
Rule 2: Be Specific, Not Vague
The Specificity Principle: The more specific your instruction, the more precise the result.
Vagueness Destroys Quality
❌ VAGUE:
"Write about productivity"
✅ SPECIFIC:
"Write a 1500-word technical blog post about productivity
for software engineers. Audience: developers aged 25-40 with
5+ years experience. Focus on: deep work, context-switching
costs, and focus techniques. Include: 3 research studies,
2-3 code examples, and 1 case study. Tone: expert but
accessible. Include SEO keywords: 'deep work', 'focus',
'developer productivity'. Format with H2 headers,
introduction, sections, and clear conclusion."
Vagueness Examples to Avoid
❌ "Check this"
✅ "Review this document for: grammar errors, logical
consistency, factual accuracy, and formatting issues.
List each error with location and suggested fix."
❌ "Find problems"
✅ "Identify and rank 5-10 critical issues that could impact
business operations. For each: describe the problem,
estimate impact (high/medium/low), and recommend solution."
❌ "Create a plan"
✅ "Create a 90-day marketing plan including: audience analysis,
4-5 channels with monthly budget allocation, weekly
tactics, success metrics, and contingency plans."
Rule 3: Assign a Clear Role
The Role Principle: Neuro AI performs better when it knows exactly what role to play.
A clear role provides:
- Expertise context
- Professional standards
- Relevant knowledge framework
- Appropriate tone and approach
Role Assignment Structure
You are [ROLE] with [EXPERIENCE] and [EXPERTISE].
Your task is to [SPECIFIC FUNCTION].
Consider [RELEVANT CONTEXT].
Strong Role Examples
For Document Analysis:
"You are a senior auditor with 15 years of experience
in financial statement audits under IFRS and local
accounting standards. You specialize in identifying
material misstatements and assessing audit risk."
"You are a management consultant specializing in
growth strategy for B2B SaaS companies. You have
experience scaling startups from $1M to $100M ARR."
"You are a regulatory compliance expert with deep
knowledge of data privacy regulations (GDPR, CCPA,
local laws). You assess legal and compliance risks."
"You are a senior software architect with 12 years
of experience designing scalable systems. You
specialize in performance optimization and AWS
infrastructure."
Rule 4: Separate Data from Instructions
The Separation Principle: Keep data and instructions distinct for clarity.
Recommended Structure
[ROLE AND CONTEXT]
DATA:
[Insert documents, tables, numbers, or information here]
TASK:
[What needs to be done with the data]
OUTPUT FORMAT:
[How the result should be presented]
CONSTRAINTS:
[Any limitations or requirements]
Structured Example
You are a financial analyst reviewing quarterly results.
DATA:
Q4 2024 Financial Summary:
- Revenue: $5.2M (vs $4.8M in Q3)
- Operating expenses: $2.1M
- Net income: $1.8M
- Customer count: 2,450 (vs 2,200 in Q3)
TASK:
1. Calculate key metrics (growth rates, margins, CAC trends)
2. Identify top 3 positive trends
3. Highlight 2-3 concerns
4. Recommend 3 specific actions
OUTPUT FORMAT:
- Executive Summary (1 paragraph)
- Metrics Dashboard (table)
- Key Findings (bullets, 3 per section)
- Recommendations (prioritized by impact)
- Appendix (supporting calculations)
CONSTRAINTS:
- Use only provided data
- Reference specific numbers in all analysis
- Keep recommendations implementable in 30 days
Rule 5: Show Examples of Desired Output
The Example Principle: Showing is more effective than telling.
When you provide examples of the desired format, Neuro AI is more likely to match that style and structure.
Few-Shot Prompting Structure
You are [ROLE].
Here are examples of the correct format:
EXAMPLE 1:
INPUT: [Sample input data]
OUTPUT: [Desired output format and style]
EXAMPLE 2:
INPUT: [Different sample data]
OUTPUT: [Matching format and style]
Now apply this approach to:
INPUT: [Your actual data]
You are a business strategist analyzing growth opportunities.
When presenting recommendations, format them like this:
EXAMPLE 1:
RECOMMENDATION: Expand sales team by 2 people
RATIONALE: Current team is at capacity; each rep manages
12 accounts, while industry standard is 8-10
EXPECTED IMPACT: +30% new customer acquisition
INVESTMENT: $250K annually
TIMELINE: Hire by end of Q1
RISK: Onboarding time may delay impact to Q2
EXAMPLE 2:
RECOMMENDATION: Improve product onboarding
RATIONALE: Customer churn in first 30 days is 18%,
above 12% benchmark
EXPECTED IMPACT: 6% improvement in retention (=$180K ARR)
INVESTMENT: $80K engineering time
TIMELINE: 8 weeks
RISK: May distract from feature roadmap
Now provide 3 similar recommendations for:
[Your business situation]
Rule 6: Use Step-by-Step Reasoning (Chain of Thought)
The Reasoning Principle: Asking Neuro AI to show its work produces better results.
Request explicit steps to ensure logical analysis and catch errors.
How to Apply
Add explicit instruction: “Solve this step-by-step. Show all intermediate reasoning and calculations.”
Example for Analytical Tasks
You are a compliance auditor checking financial controls.
TASK: Assess whether the company's bad debt reserve is adequate.
DATA:
Accounts receivable aging:
- 0-30 days: $2,000K (customers: solid credit)
- 31-60 days: $800K (customers: mixed credit)
- 61-90 days: $400K (customers: weak credit, payment delays)
- 90+ days: $600K (customers: bankruptcy risk)
INSTRUCTIONS - WORK STEP BY STEP:
1. For each aging bucket, estimate probability of collection
2. Calculate required reserve for each bucket
3. Calculate total required reserve
4. Compare to company's existing reserve
5. State whether reserve is adequate, excessive, or deficient
6. Show all intermediate calculations
7. Explain your assumptions for each bucket
FORMAT:
- Aging Bucket Analysis (table with probabilities)
- Reserve Calculation (step-by-step math)
- Assessment (adequate/deficient by $X)
- Recommendations
The Format Principle: Clear output specifications prevent wasted revision rounds.
Neuro AI works best when you specify exactly how you want results formatted.
✅ Output medium: (Excel, Word document, HTML, Markdown, etc.)
✅ Structure: (Executive summary + details, sections, tables)
✅ Length: (1 page, 5 pages, word count)
✅ Detail level: (High-level overview vs detailed analysis)
✅ Visual elements: (Charts, diagrams, tables)
✅ Language and tone: (Technical, business casual, formal)
✅ Specific sections: (What must be included)
Example Specification
OUTPUT FORMAT SPECIFICATIONS:
Medium: PDF-ready document
Structure:
1. Executive Summary (1 page max)
2. Key Findings (bullets, 5-7 items)
3. Detailed Analysis (3-5 sections with subheadings)
4. Recommendations (table with: # | Action | Impact | Timeline | Cost)
5. Appendix (supporting data, calculations)
Length: 8-12 pages total
Visuals: 4-6 professional charts/tables
Tone: Professional but conversational
Font specs: Heading style shows importance hierarchy
Emphasis: Bold for key numbers, italics for concepts
Specific requirements:
- All claims backed by data citations
- ROI calculation for each recommendation
- Risk assessment for each major recommendation
Rule 8: Set Quality Standards Explicitly
The Quality Principle: Define what “good” means for your specific task.
Without quality standards, Neuro AI might produce generic results that don’t meet your needs.
Quality Standards to Specify
✅ Accuracy: How precise must answers be?
✅ Completeness: What aspects must be covered?
✅ Depth: Surface level or deep analysis?
✅ Actionability: How specific/implementable?
✅ Credibility: Must sources be cited?
✅ Originality: Can it be generic or must it be custom?
Example Quality Standards
QUALITY STANDARDS FOR THIS ANALYSIS:
ACCURACY:
- All numbers must match source documents exactly
- No rounding differences > $1K
- Calculations must be verified
COMPLETENESS:
- Must address all 5 criteria mentioned
- No key aspect can be omitted
- Must include risk assessment for each finding
DEPTH:
- Analysis must go beyond surface observations
- Include root cause analysis, not just symptoms
- Explain the "why" behind each finding
ACTIONABILITY:
- Recommendations must be specific enough to implement
- Include step-by-step instructions
- Specify responsible party and timeline for each action
CREDIBILITY:
- All claims backed by data citations
- When using industry benchmarks, cite the source
- Distinguish between facts, analysis, and assumptions
ORIGINALITY:
- Tailor analysis to OUR business context
- Don't provide generic recommendations
- Address our specific constraints and opportunities
Part 2: Advanced Prompting Techniques
Rule 9: Protect Against Hallucinations
The Accuracy Principle: Neuro AI can sometimes generate false information. Protect yourself with explicit safeguards.
Technique 1: Require Citations
INSTRUCTION:
"When answering, you MUST cite specific sources. If information
comes from the document, write: 'According to [document name],
page X: [quote]'. If information is NOT in the documents,
write 'This information is not provided in the documents.'
NEVER make up information or use general knowledge beyond
what's provided."
Technique 2: Limit Knowledge Domain
INSTRUCTION:
"Use ONLY the information provided in the attached documents.
Do not use general knowledge about [topic]. If something isn't
in the documents, say so explicitly. Do not infer, assume, or
apply external knowledge."
Technique 3: Require Confidence Levels
INSTRUCTION:
"For each conclusion, specify your confidence level:
- CONFIDENT: Information is explicitly stated in documents
- INFERRED: Logical conclusion from information provided
- UNCERTAIN: Limited information, could be wrong
- UNAVAILABLE: No information provided
Example:
Finding: Revenue will grow 20% next quarter
Confidence: INFERRED (based on current growth rate and
Q4 pipeline data)"
Rule 10: Use Comparative Analysis
The Clarity Principle: Comparing options helps clarify thinking and identify best choices.
Comparative Analysis Structure
TASK: Compare [Option A], [Option B], and [Option C]
COMPARISON DIMENSIONS:
1. [Dimension 1] - Why it matters
2. [Dimension 2] - Why it matters
3. [Dimension 3] - Why it matters
For each comparison:
- Describe pros and cons
- Assign a score (1-5)
- Explain the score
- Note any caveats or context
SUMMARY:
- Overall winner and why
- When Option B might be preferred
- When Option C makes sense
- Key decision factors
Rule 11: Request Devil’s Advocate Analysis
The Challenge Principle: Ask Neuro AI to argue against the proposal to identify risks and weaknesses.
INSTRUCTION:
"I've proposed the following strategy: [Strategy description]
Now I want you to argue AGAINST this strategy. Provide:
1. The 5 strongest arguments against this approach
2. Real-world evidence that it might fail
3. What could go wrong
4. Hidden assumptions I'm making
5. What would need to change for you to recommend this
Be thorough and compelling in your counter-arguments."
Rule 12: Apply Constraint-Based Thinking
The Creativity Principle: Tight constraints force innovative solutions.
TASK: We need to increase revenue 30%, but we CANNOT:
- Raise prices
- Add features (locked roadmap)
- Hire more people
- Increase marketing spend
CONSTRAINT: We only have $50K/month budget
Given these constraints, what are 5 creative ways we could grow?
For each:
- Why it could work
- What needs to happen
- Timeline and resources
- Probability of success
Part 3: Practical Application Templates
Template 1: Analysis Task
You are [ROLE] with expertise in [DOMAIN].
DATA:
[Provide all relevant information, documents, numbers]
TASK - Analyze for:
1. [Key aspect 1]
2. [Key aspect 2]
3. [Key aspect 3]
SPECIFIC REQUIREMENTS:
- Quantify findings where possible
- Reference specific data points
- Explain significance of findings
- Identify gaps or inconsistencies
OUTPUT FORMAT:
- Summary (1 paragraph)
- Key Findings (table or bullets)
- Detailed Analysis (by section)
- Recommendations (ranked by impact)
- Appendix (supporting details)
QUALITY STANDARDS:
- All numbers verified against source data
- All claims backed by evidence
- Distinguish facts from interpretations
- Consider alternative explanations
Template 2: Writing/Content Creation
You are [ROLE] specializing in [EXPERTISE].
PURPOSE:
[Why this content is needed and its business value]
AUDIENCE:
[Who will read this - demographics, knowledge level,
what they care about, pain points]
TONE & VOICE:
[Professional/casual/expert/friendly, formal/informal]
FORMAT:
[Document type, length, structure, visual elements]
KEY MESSAGES:
[Top 3-5 main points that must be conveyed]
MUST INCLUDE:
[Specific content, examples, or elements that are required]
MUST AVOID:
[Topics, language, or approaches to exclude]
CALL TO ACTION:
[What should the reader do after reading]
SUCCESS CRITERIA:
[What makes this piece successful]
Template 3: Strategy/Planning
You are [ROLE] with [EXPERIENCE].
CURRENT SITUATION:
[Where we are now - metrics, challenges, context]
DESIRED SITUATION:
[Where we want to be - goals and success criteria]
CONSTRAINTS:
[Budget, timeline, resources, organizational limits]
TIMELINE:
[How long we have to execute this strategy]
DEVELOP:
1. Situation analysis [What's driving this need?]
2. Strategic options [4-6 alternatives]
3. Evaluation [Pros/cons of each option]
4. Recommendation [Top choice with rationale]
5. Implementation plan [Step-by-step with owners, timeline]
6. Success metrics [How we'll measure success]
7. Risk assessment [What could go wrong and mitigation]
OUTPUT:
[Format, length, who needs to see this]
Template 4: Problem Solving
You are [ROLE] specializing in [DOMAIN].
PROBLEM STATEMENT:
[Clear, specific description of the problem]
CONTEXT:
[Background information, why it matters, constraints]
SUCCESS DEFINITION:
[What solved looks like - specific, measurable]
REQUESTED ANALYSIS:
1. Root cause diagnosis [What's actually driving this?]
2. Solution options [4-6 possible approaches]
3. Evaluation [Pros/cons, implementation difficulty, cost]
4. Top recommendation [Which solution and why]
5. Implementation steps [Detailed execution plan]
6. Success metrics [How to measure improvement]
7. Timeline [When should results be visible]
CONSTRAINTS TO CONSIDER:
[Budget, timeline, team capacity, other limitations]
AVOID:
[Surface solutions, quick fixes, Band-Aid approaches]
Part 4: Common Mistakes to Avoid
Mistake 1: Mixing Data and Instructions
❌ WRONG:
"Analyze the Q4 sales report which shows revenue of $5.2M
and customer growth of 15% and please tell me if that's good
and what we should do about it and format it nicely."
DATA:
Q4 Results:
- Revenue: $5.2M (up from $4.8M in Q3)
- Customer count: up 15% YoY
- Margin: 35%
TASK:
1. Assess performance vs benchmarks
2. Identify top 3 trends
3. Recommend 3 actions
FORMAT:
Executive summary + detailed analysis + recommendations
Mistake 2: Vague Success Criteria
❌ WRONG:
"Give me a report on our market."
"Create a market analysis report covering:
1. Market size and growth (with projections)
2. Top 10 competitors (with positioning)
3. Customer segments (demographics and needs)
4. Entry barriers and competitive advantages
5. Growth opportunities for our company
Include: market research sources, competitive positioning
map, TAM/SAM/SOM analysis, and 3 strategic recommendations."
Mistake 3: Assuming Neuro AI Knows Your Context
❌ WRONG:
"Update the pricing strategy based on recent changes."
"Our pricing strategy context:
- Current price: $99/month (vs $89 for Competitor A, $129 for Competitor B)
- Customer segment: Mid-market B2B SaaS
- Recent changes: 3 new competitors entered market, our churn up 2%
- Goal: Increase revenue 15% without increasing churn
Analyze whether we should:
1. Increase price to $129
2. Keep current pricing
3. Introduce tiered pricing
For each option: impact on revenue, churn, positioning,
and recommended messaging."
Mistake 4: Not Being Specific About Constraints
❌ WRONG:
"Create a marketing plan."
"Create a 90-day marketing plan with these constraints:
- Budget: $50K
- Team: 1 marketing manager + 1 contractor (10 hrs/week)
- Timeline: Must show results by end of Q1
- Goals: 25% increase in website traffic, 100 qualified leads
- Channels: Focus on LinkedIn, content, and partnerships
(not paid ads or traditional marketing)
- Audience: B2B enterprise software buyers, $500M+ companies
- Current situation: 500 website visitors/month, no leads yet"
Part 5: Final Checklist
Before submitting any prompt, verify:
Content Checklist
□ Is the role/context clear?
□ Are all necessary data points included?
□ Is the task specific (not vague)?
□ Is the format clearly specified?
□ Are constraints documented?
□ Have I provided examples?
□ Are quality standards explicit?
□ Did I ask for step-by-step reasoning if needed?
□ Did I include safeguards against hallucinations?
Clarity Checklist
□ Could someone unfamiliar with this understand it?
□ Are all abbreviations/jargon explained?
□ Is the structure logical and easy to follow?
□ Did I use specific verbs (not vague ones)?
□ Are numbers and quantities exact (not approximate)?
□ Would anyone misinterpret this?
Completeness Checklist
□ Is all necessary context provided?
□ Are there any implicit assumptions I should state?
□ Did I specify success criteria?
□ Are there any edge cases I should mention?
□ Did I explain why this matters?
Summary: The 12 Main Rules
- Use the three-part structure — Role + Task + Format
- Be specific, not vague — Details drive quality
- Assign a clear role — Provide expertise context
- Separate data from instructions — Structure for clarity
- Show examples of desired output — Few-shot prompting
- Use step-by-step reasoning — Chain of thought
- Specify expected output format — Prevent revisions
- Set quality standards explicitly — Define “good”
- Protect against hallucinations — Require citations and limits
- Use comparative analysis — Compare options clearly
- Request devil’s advocate — Identify risks and weaknesses
- Apply constraint thinking — Force creative solutions
Clear Role + Specific Task + Exact Format + Quality Standards
+ Data Separation + Step-by-Step = Excellent Results
Remember
Every minute spent perfecting your prompt saves multiple minutes of revision and iteration.
Invest in clarity upfront. Reap the benefits in results quality.
Neuro AI Main Prompting Rules
Master these rules. Get consistently excellent results.
Last updated: January 2026
Status: Essential Reference
Use this document as your prompting standard.