2 AI for Problem Solving from Slides

• Personas, problems, pains, & gains emphasize people over technologies.
• It's important to know what problem is being solved and for whom.

ComfortAlble Scenario

• ComfortAlble is an Al powered smart home system.
• It simplifies smart home use for its users.
• Enhances tenant-domicile automation experience.
• Reduces management complexity & maintenance costs.
• Tackles tactical challenges with strategic Al solutions.
• Scales and improves smart home usability.

Personas

• Personas define the "who".
• They are composite profiles of groups of persons.
• They act as a reference point providing context for users and buyers.
• For example, "Charlie the Chocoholic" is a 10-year-old boy who seeks chocolate that delivers adventure and guilt-free pleasure with new, imaginative, healthier options.
• User personas focus on usage behavior, pains & goals, UX preferences, behavioral insights, empathy & understanding, and qualitative feedback.
• Buyer personas focus on buying motivation, budgetary constraints, business goals & pain points, value proposition & benefits, external factors of impact, and quantitative data.

Agenda includes study of

• Personas
• Problems
• Prompt engineering
• Jobs-to-be-done

Generative Al & Problems and Personas

• An exploration of how Generative Al can be used in problems and personas
• GAI is creative branch of AI.
• GAI is made up of multiple AI like deep learning, neural networks, & NLP.
• It makes new things like text, images, and music
• It also makes things up, like hallucinations

Al Chatbot Dialog

• Understood in considering the 3 roles of User, Assistant, & System
• User is the person requesting
• Assistant supplies the Al interface providing responses
• System governs Assistant behavior

Prompt Engineering Essentials

• Be exact with request.
• Request an expert-level role if needed.
• Show preferred format.
• Use delimiters & data for request clarity.
• Provide step-by-step instructions.
• Give examples to improve understanding.
• Quality input creates quality output.
• Models can't read minds, brief replies lack depth, jumbled input equals jumbled output and no examples equals no clarity.

Session Context

• Let AI know who you are, who your audience is, your goals, the task, constraints, and desired formats.
• More context generally decreases AI hallucinations.

Session Starter Prompt Considerations

• Persona description, role, and challenges.
• Outcomes.
• Tone.
• What? It sets roles, context and tone.
• Why? It keeps the session aligned on outcomes.
• Reusable & Shareable: Keeps team sessions consistent.
• Focus on Problems: Start with problems, not solutions.

Picking a Persona

• Choose a problem statement and a customer persona to guide exploration of potential solutions.
• Gen Al session starter should include structured date and external sources to be used in class exercises, using tools like Mural & Gen Al Chatbot.

Session Context Window

• Setting an initial context helps future responses
• Managing context will avoid forgetting earlier context
• Accuracy affected when the context window slides from response.

Adding Outside Smarts

• Use Al-assisted search tools like Copilot or Perplexity for near-real-time, relevant data.
• Upload documents (PDF, PNG, etc.) for domain-specific knowledge.
• This improves focus on relevant, accurate data by ensuring AI works with current, precise context.

Leveraging Structured Data

• Think beyond Markdown and use formats such as e YAML, JSON, & CSV
• Reduces vagueness and guarantees clearer, more relevant responses; structured data helps teams keep context.

Prompts by Reference

• Ideas explored with a Teresa Torres Opportunity Solution Tree, customer pains and gains mapped, key interactions visualized, product positioning defined using Geoffrey Moore's framework, and testable hypotheses framed.

Prompts by Template

• Template created for clarity, consistency, and collaboration
• Teams align with the same framework
• Can adapt with quality

Framing the Problem

• Important to articulate underserved needs
• Can lead to problem statements

Framing the Problem Information

• Use a statement to know what was learned in problem space
• Keeps efforts focused
• Keep up to date with the costumer's story for stakeholders
• Allows for prioritization with engineering

Problem Frame Example

• "I am a security officer who is trying to ensure efficiency but passengers arrive unprepared, because they lack awareness, which makes me feel frustrated."

When Prompting

• What are the key pain points for the person?
• What outcomes is the persona trying to achieve?
• What obstacles are in the way of people succeeding?
• How does that persona feel about the issue?

Activities include

• Personas: Pains, Gains, & JTBD to identify with the persona
• Frame the Problem by documenting one under an investigation

Rules of the Road

• DO establish context in new sessions, explore problems before solutions, iteratively evolve the session, generate conversation for humans & engage ethically.
• DON'T start prompting without a plan, prompt using solution speak, ask for features with no context, craft tactical plan from scratch & ignore your organization's Al & Privacy policies.