AI Best Practices for Business
Tips and techniques to get the most from your AI workspace.
Prompting Fundamentals
The PREP Framework
Write better prompts every time:
| Element | Description | Example |
|---|---|---|
| Purpose | What do you want? | "Write an email..." |
| Role | Who should AI be? | "...as a professional marketer..." |
| Expectations | Format, length, tone | "...in 3 paragraphs, friendly tone..." |
| Provide | Give context | "...about our new product launch." |
Good vs. Bad Prompts
| Bad Prompt | Good Prompt |
|---|---|
| "Write an email" | "Write a follow-up email to a potential customer who attended our webinar but didn't schedule a call. Keep it under 150 words, friendly but professional, with a clear call-to-action to book a meeting." |
| "Make this better" | "Improve this paragraph by making it more concise, using active voice, and emphasizing the cost savings for the customer." |
| "Help me with marketing" | "Create 5 LinkedIn post ideas for our B2B software product that target HR managers dealing with employee retention challenges." |
Prompt Templates
Keep a library of prompts that work:
Email Template:
Write a [type] email to [recipient].
Purpose: [goal]
Key points: [bullets]
Tone: [voice]
Length: [word count]
End with: [CTA]
Content Template:
Create [content type] about [topic].
Target audience: [who]
Key message: [main point]
Format: [structure]
Include: [specific elements]
Voice: [brand voice]
Organization Strategies
Project Structure
Keep your workspace organized:
✅ Good Structure
├── Marketing
│ ├── Clear project description
│ ├── Relevant documents only
│ └── Active threads organized
├── Operations
├── Sales
└── Finance
❌ Bad Structure
├── Stuff
├── More stuff
├── Random
└── Everything else
Naming Conventions
| Type | Good Example | Bad Example |
|---|---|---|
| Project | "Marketing - Q1 2024" | "Marketing stuff" |
| Thread | "Blog post - AI trends article" | "writing" |
| Document | "Brand Voice Guidelines v2" | "doc1" |
| GPT | "Customer Response Assistant" | "helper" |
Regular Maintenance
| Task | Frequency | Why |
|---|---|---|
| Archive old threads | Weekly | Keep workspace clean |
| Update documents | Monthly | Keep AI informed |
| Review GPT performance | Monthly | Improve outputs |
| Remove unused projects | Quarterly | Reduce clutter |
GPT Optimization
Writing Good Instructions
Your GPT instructions should include:
| Component | Purpose | Example |
|---|---|---|
| Role definition | Who the GPT is | "You are a professional copywriter for a B2B software company" |
| Guidelines | How to behave | "Always use active voice, avoid jargon" |
| Constraints | What to avoid | "Never make claims without evidence" |
| Format | Output structure | "Provide 3 options, each under 100 words" |
| Examples | Reference outputs | "Like this example: [paste sample]" |
Iterative Improvement
When outputs aren't perfect:
- Identify the issue - What's wrong specifically?
- Update instructions - Add guidance to prevent it
- Test again - Try similar prompts
- Repeat - Keep refining until consistent
Common GPT Fixes
| Issue | Solution |
|---|---|
| Too generic | Add more specific examples |
| Wrong tone | Include voice description with adjectives |
| Too long | Add word/sentence limits |
| Missing elements | List required components |
| Inconsistent | Add "always" and "never" rules |
Document Management
What to Upload
| Priority | Document Type | Impact |
|---|---|---|
| High | Company overview | Every response |
| High | Product/service details | Accuracy |
| High | Brand guidelines | Consistency |
| Medium | Process documents | Reference |
| Medium | Templates | Formatting |
| Lower | Historical data | Context |
Document Quality
Better documents = better AI outputs:
| Good | Bad |
|---|---|
| Clear headings | Wall of text |
| Bullet points | Run-on paragraphs |
| Current information | Outdated content |
| Complete context | Fragments |
When to Update
Update documents when:
- Information changes (pricing, features, team)
- AI outputs become incorrect
- New products/services launch
- Processes change
- Quarterly review reveals gaps
Team Usage
Training Your Team
| Training Element | Time | Focus |
|---|---|---|
| Platform basics | 30 min | Navigation, core features |
| Department GPTs | 30 min | Specific tools for their role |
| Prompting skills | 30 min | Writing effective prompts |
| Q&A | 30 min | Address specific questions |
Usage Guidelines
Establish clear guidelines:
| Guideline | Reason |
|---|---|
| Always review AI outputs | Catch errors before use |
| Use appropriate projects | Keep context organized |
| Report issues | Improve GPTs over time |
| Share successful prompts | Help teammates |
| Don't upload sensitive data | Maintain security |
Quality Control
| Task | Who | When |
|---|---|---|
| Review AI outputs | User | Every time |
| Spot-check quality | Lead | Weekly |
| GPT performance review | Admin | Monthly |
| User feedback collection | Admin | Quarterly |
Troubleshooting
Quick Fixes
| Problem | Try This |
|---|---|
| AI seems confused | Start a new thread with clear context |
| Wrong format | Be more specific about desired format |
| Missing information | Check if relevant docs are uploaded |
| Inconsistent outputs | Review and update GPT instructions |
| Slow responses | Use a faster model for simple tasks |
When to Get Help
Contact support when:
- Technical issues with the platform
- Persistent quality problems
- Questions about best practices
- Need for advanced customization
- Security or privacy concerns
Advanced Techniques
Chain Prompting
Break complex tasks into steps:
Step 1: "Analyze this customer feedback and list the top 5 themes"
Step 2: "For each theme, suggest 2 improvements we could make"
Step 3: "Prioritize these improvements by impact and effort"
Step 4: "Draft a summary for the leadership team"
Context Stacking
Build context through conversation:
You: "I'm working on our Q2 marketing plan. Our focus is small business customers in professional services."
AI: [acknowledges]
You: "We have $10,000 budget and 3 months."
AI: [acknowledges]
You: "Now create a channel strategy with this context."
AI: [responds with full context]
Output Refinement
Iterate to perfection:
You: "Write a product description"
AI: [first draft]
You: "Make it shorter"
AI: [shorter version]
You: "Add more benefits"
AI: [with benefits]
You: "Match our casual brand voice"
AI: [final version]
Continuous Improvement
The best AI users constantly refine their approach. Save prompts that work, update GPTs based on results, and share successes with your team.