Workshop 2: Identifying High-Impact AI Use Cases in Your Workflow
Welcome to Week 2 of our AI integration workshop! After establishing foundations in Week 1, we're now focusing on identifying and implementing high-impact AI use cases in your professional workflow. Today, we'll explore how to select the right tasks for AI assistance, evaluate their potential impact, and develop effective prompts to maximize your productivity.
By the end of this session, you'll have practical frameworks to transform your repetitive tasks into streamlined AI-powered processes, saving you valuable time and mental energy for more creative and strategic work.

by Haojun See

On The Ground - AI Adoption for Businesses and Individuals
Why Use Case Selection Matters
Amplification, Not Transformation
AI doesn't fix bad workflows; it amplifies what you give it. When you provide clear inputs and well-defined expectations, AI can deliver remarkable results. However, applying AI to flawed processes only accelerates those flaws.
Preventing Wasted Effort
Use case clarity prevents wasted effort and frustration. By identifying appropriate tasks, you avoid the common pitfall of trying to force AI into workflows where it adds little value or creates more work than it saves.
Starting Point
Think: "What task annoys you weekly but follows a pattern?" These repetitive, predictable tasks with clear inputs and outputs make excellent candidates for AI assistance.
Impact vs Effort Matrix
High Impact / Low Effort
The sweet spot for AI implementation. These tasks offer significant time savings or quality improvements while requiring minimal setup and maintenance. Prioritize these opportunities first.
High Impact / High Effort
Plan for later implementation. These tasks could provide substantial benefits but require significant investment in preparation, training, or oversight. Schedule these for future phases.
Low Impact / Low Effort
Optional improvements. While these tasks are easy to implement, they offer minimal benefits. Consider implementing if time allows after addressing higher-priority items.
Low Impact / High Effort
Avoid these tasks. The return on investment simply isn't worth it. Focus your energy elsewhere rather than expending significant effort for minimal gains.
Characteristics of AI-Compatible Tasks
Text-heavy inputs
Tasks involving substantial text processing, such as emails, meeting transcripts, standard operating procedures, or client briefs, are ideal candidates for AI assistance.
Repeatable formats
Work that follows consistent structures like regular reports, status updates, or standardized lists can be effectively templated and automated with AI tools.
Objective goals
Tasks with clear objectives such as summarizing content, rewriting for different audiences, extracting key information, or structuring data yield the best results.
Low judgment steps
Processes that don't require complex decision-making, legal approval, or nuanced ethical considerations are safer and more reliable when automated.
Today's Goal
Break Down Your Tasks
Learn to dissect your real-world workflows into AI-friendly components. Identify which steps can be automated and which require human oversight.
Build Effective Prompts
Develop and refine prompts around one strong use case from your workflow. Practice the art of clear instruction to get consistent, high-quality results.
Learn From Examples
Study successful and unsuccessful AI implementations to understand what works, what doesn't, and why. Apply these insights to your own use cases.
Real Examples – Legal Sector
Case Notes to Client Updates
Transform detailed technical case notes into clear, accessible client updates. AI can maintain accuracy while translating legal terminology into language clients can easily understand.
Contract Analysis
Extract critical dates, obligations, and risk factors from lengthy contracts. AI can quickly identify and organize key information that would take hours to manually compile.
Plain Language Translation
Rewrite complex legalese into clear, plain English for non-specialist audiences. AI excels at maintaining meaning while simplifying language and structure.
Real Examples – SME Sector
Small and medium enterprises can leverage AI for numerous everyday tasks. Policy emails can be rewritten for specific audiences like part-time staff. Onboarding survey results can be quickly summarized to identify patterns and actionable insights. Rough bullet points can be transformed into polished proposal introductions. Even informal WhatsApp group discussions can be converted into professional client updates, saving valuable time while maintaining communication quality.
Examples that Are NOT Ideal
Multi-party Negotiations
These interactions are typically unstructured and dynamic, requiring nuanced human judgment and real-time adaptation to changing positions and emotional cues.
Sensitive Compliance Reviews
The risk of AI hallucination or misinterpretation is too high for critical compliance matters where errors could have serious legal or regulatory consequences.
Legal Opinions
Tasks requiring authoritative legal judgment, factual guarantees, or professional liability should remain human-driven to ensure accuracy and appropriate context.
Case Study: The 6-Minute Email
Input
Raw, unstructured Zoom meeting notes from a detailed legal consultation call, containing technical terminology and client questions.
Prompt
"Summarize these notes into a 3-point email with a friendly, professional tone that addresses the client's main concerns."
Result
A clear, concise client communication that took only 6 minutes to produce (75% faster than manual drafting), with fewer requests for clarification.
Group Question
Would you implement this in your workflow? What quality checks would you add to ensure accuracy and appropriateness?
Workflow Dissection Framework
This framework helps you break down your existing workflows and identify opportunities for AI-powered automation. Let's dive in step-by-step:
Trigger
The event that initiates the workflow. This could be a client request, a scheduled report, or a team meeting - anything that kicks off the process.
Input
The raw materials you work with to complete the workflow.
This is the information, data, or assets you need to get started.
Use Case Canvas Template
Live Prompting Practice

Select a Task
Choose one workflow from your use case canvas
Structure Your Prompt
Follow the recommended format for clarity
Test and Refine
Evaluate results and adjust as needed
For effective prompts, follow this structure: Begin with "You are a [relevant role]" to set context.
Clearly state "Your job is to [specific task]" to define the objective.
Specify the "Input: [type]" you're providing.
Finally, detail the "Output: [format, tone, level of detail]"
Demo – Prompt Evolution
Weak Prompt
"Make this better."
Too vague, providing no guidance on what "better" means or what output is expected. Likely to produce inconsistent results that may not address your needs.
Better Prompt
"Rewrite this in a professional tone for a confused client."
Offers more direction regarding tone and audience, but still lacks specificity about format and structure.
Best Prompt
"You are a legal assistant. Rephrase this update into a 3-point email that's empathetic, concise, and accurate."
Provides clear role context, specific output format, desired tone, and length parameters for optimal results.
Prompt Variants by Function
Rewrite
"You are a client communication specialist. Turn this technical note into a plain English update that a non-expert can understand. Maintain all key information while using accessible language and a friendly, reassuring tone."
Extract
"You are a risk assessment analyst. Review this contract and list all key risks, deadlines, and next steps in a bulleted format. Organize by priority (high to low) and include relevant clause references."
Convert
"You are a project manager. Transform these meeting notes into a clear, actionable checklist for my team. Include responsible parties, deadlines, and prioritize items by urgency."
Prompting Mistakes to Avoid
The most common prompting errors fall into three categories.
Vague instructions like "Make this sound good" provide insufficient guidance for the AI to deliver useful results.
Overly ambitious requests that try to accomplish too much in one prompt often yield incomplete or superficial responses.
Perhaps most problematic are prompts asking for opinions or decisions on matters requiring professional judgment, like "Should I approve this?" Remember that AI should inform your decisions, not make them for you.
Peer Feedback (Live Activity)
1
Partner Up
Form pairs with someone in a similar field or role to maximize relevant feedback.
Exchange your use case descriptions and prompt drafts.
2
Apply the Rubric
Evaluate your partner's prompt using the structured rubric: Is the task clearly defined? Is the desired output format specified? Does it follow the recommended structure with role, task, format, and style? What risks or limitations should they consider?
3
Provide Constructive Feedback
Offer specific suggestions to strengthen their prompt.
Focus on clarity, specificity, and potential edge cases they might not have considered.
4
Revise and Share
Take 5 minutes to incorporate feedback and improve your prompt. Share your revised version with the larger group if time permits.
Mini-Sprint Briefing
Select a Real-World Task
Choose one actual task from your current workload that meets our AI-compatibility criteria. Pick something meaningful that you'd genuinely like to optimize.
Create Prompt Variations
Develop three slightly different versions of your prompt. Make small, deliberate changes to test specific variables like tone, format, or level of detail.
Compare Outputs
Run all three prompts and carefully compare the results. Note which elements were most effective and which might need further refinement.
Document Your Findings
Write a brief reflection addressing: Which prompt worked best and why? What surprised you about the results? Would you implement this in your actual workflow?
Iteration Template
Recap – What Makes a Use Case 'Worth It'
4+
Weekly Frequency
Tasks that occur at least several times per week offer the best return on your investment in prompt development and refinement.
75%
Time Savings
Effective AI implementation should reduce the time spent on a task by a significant margin while maintaining or improving quality.
15min
Setup Time
The ideal AI use case requires minimal preparation time, with prompt development and testing completed in under 15 minutes.
As you reflect on today's session, consider: What's one task you'll test AI on this week?
What preparation do you need (templates, cleaner data, clearer formats)?
And finally, what task did you initially think was suitable for AI but now realize requires too much human judgment?
Contact us with questions at [email protected] or via WhatsApp at wa.me/6598419481.