Step 1 of 7: From Idea to Impact

🎁 Free: The Core Framework

Pain × Frequency × Feasibility

Before you touch a prompt or a model, answer three questions about the problem, not the AI:

1. Pain: how much does this actually hurt the user today?
2. Frequency: how often do they hit this pain?
3. Feasibility: can AI meaningfully close the gap, or is this a UX problem wearing an AI costume?

Most failed AI features skip straight to "what can the model do" instead of starting here. If you can't score all three high, you don't have an AI opportunity. You have a science project.

🔓 Unlocked: Sections 1-3 in Full

1. The 4-Source Research Sweep, with actual execution steps

Market/competitor scan, copy-paste Perplexity prompt:

"I'm building [feature type] for [user type]. List every company shipping something similar in the last 12 months. For each: what they built, what's public about their approach (RAG/fine-tune/agent), what users say in reviews/Reddit/Twitter about where it falls short. Prioritize failure reports over marketing claims."

User interview patterns specific to AI: standard questions ("what's frustrating about X") don't work because users haven't imagined the AI version yet:

• Ask about the manual workaround they currently use, in detail. The workaround IS the spec
• Ask "if a smart intern did this for you, what would you still double-check?" This surfaces the trust boundary before you've built anything
• Never ask "would you use an AI that does X." Hypothetical AI questions get uniformly positive, useless answers

Support ticket mining: pull the last 90 days, cluster by root cause (not by category tag, categories lie), and flag any ticket where a human agent's answer was templated/repetitive. Templated answers are your highest-confidence AI candidates because you already have proof the pattern is stable enough to automate.

Shadow AI detection: check whether people are pasting your product's data into ChatGPT already. Ask support/sales/CS directly: "has a customer ever mentioned using AI tools alongside our product to solve X?" This is the single strongest demand signal available, and most teams never ask the question.