Gains:
- Ability to understand problem-oriented thinking, not solution-oriented thinking, and use artificial intelligence to explore problem areas and generate ideas
- Ability to distinguish whether a problem is 'vitamin or painkiller' and hypothesize with artificial intelligence
- Being able to understand that the list of ideas produced by artificial intelligence is a starting point and that real pain can only be verified in the field.
Most startups die not because of a bad idea, but because they made a product for a "problem" that no one really wanted solved. Startup founders often start with a solution: “I'm going to make an app, this is how it will work.” Experienced founders, however, start with a problem: “What real pain are people experiencing right now, and are they willing to pay for it?” In this unit we will learn to use AI (artificial intelligence — software that generates ideas and text) in the right order, that is, first for problem discovery. Let's set the limit from the beginning: AI can generate hundreds of ideas, but only people in the field can tell which pain is real.
Thinking problem-oriented, not solution-oriented
A problem space is the set of difficulties that people encounter while doing their job or living. Solution space is the products and services produced to solve these difficulties. The rookie mistake is to jump straight into the solution area; Because imagining the solution is fun, understanding the problem is boring and troublesome. However, the perfect solution to the wrong problem is garbage.
A good problem has three characteristics. It occurs frequently (nobody makes a habit of a rare problem). It is painful (one does not exert effort for mild discomfort). Expensive (a solution is valued if the problem costs time, money or reputation). You can use AI to “dig” problems in these three axes.
Vitamin or painkiller?
It is one of the most useful distinctions in entrepreneurship. Painkillers solve real and immediate pain; Otherwise, a person cannot do his job, so the willingness to pay is high (for example, a collection system for a tradesman who cannot work without being paid). Vitamins, on the other hand, are a pleasant but dispensable improvement; "It would be nice to have", but life goes on without it (for example, a colorful motivational app). Vitamins can make money, but it's a much tougher road. When you give the AI an idea, it immediately asks "is this a painkiller or a vitamin, what is the evidence?" Make them question.
Tip: The quickest way to find out if a problem is a painkiller is to ask: “What are people doing right now to solve this problem and how much time/money are they spending on it?” If money/effort is currently being spent on a solution (even if it is bad), the pain is real.
Step by step: Problem discovery with AI
- Select a people/segment. Think for a specific group, not “everyone” (e.g. “solo dentists”).
- Scratch your days. Extract from the AI a typical day for that group, their recurring frustrations, and their current coping strategies.
- Turn problems into hypotheses. Turn each hardship into a testable sentence: “Group X suffers Z in situation Y.”
- Vitamin/pain reliever filter. Evaluate each hypothesis on this axis; Eliminate the weak.
- Generate an evidence question. For every strong hypothesis “how do I test this in a week?” Answer the question.
- Take it to the field. The ultimate filter is talking to a real human, not an AI (next unit).
Note: The problem list that the AI produces is a "starting point", not reality. AI generalizes; Your segment's true pain may not be on the list, or an item on the list may not apply to you.
three mini cases
Case 1 — Turning from solution to problem. A founder wanted to make “an AI recipe app” (he started with the solution). When he had the AI dig into the day of his target audience (new parents), he found that the real pain was not the recipe, but "eating quickly from whatever was on hand" and "lack of time." He redefined the problem: "less material, less time". The product idea has changed radically and has become much clearer. Ultimately, he spoke to 15 people; 12 confirmed this pain.
Case 2 — Eliminating the vitamin too early. One team was considering an app that “sends a daily inspirational quote to employees.” When you get the AI to test for vitamins/painkillers and ask “what are people spending on this now?” When you looked at the question honestly, the answer was "nothing." No one was spending money/effort for a word of inspiration. They dropped the idea early; They developed products for months and then got away with learning. The cheapest failure is the one that is noticed before you even start.
Case 3 — Secret painkiller. One founder focused on small tour companies. AI made this group's day; There were many problems, but one stood out: constant double bookings and lost money due to WhatsApp and manual tracking of reservations and collections. This was a frequent, painful and expensive problem—the classic painkiller. He abandoned the other 6 ideas and focused on this one pain.
Four copyable templates
1) Digging a day:
Your role: customer discovery specialist. Target group: [e.g. single working dentist].Describe, hour by hour, a typical workday for this person. List the recurring problems you encounter at each stage and how you are currently dealing with them (current solution/remedy). Where you are unsure, write "assumption, must be verified". Remember that you are generalizing.
2) Turning the problem into a hypothesis:
Take the following list of pain points and turn each one into this pattern: "[segment] is experiencing [pain] at the moment of [situation], resulting in [cost]." Then mark each hypothesis as "painkiller/vitamin" and write the rationale in one sentence.
3) Vitamin/painkiller stress test:
Evaluate the following problem hypothesis: [hypothesis]. Score 1-5 on the axes of (1) frequency, (2) pain intensity, (3) cost. (4) Answer the question "What are people doing and spending for this problem right now?" (5) Give a total pain relief score and, if poor, explain why it is poor.
4) Cheap test design:
I want to test the following hypothesis in a week, without writing code: [hypothesis]. Suggest me 3 inexpensive testing methods (e.g. interview, pre-registration page, mock door test). For each: what do I measure, how many people are enough, which result says "the bitter truth" which result says "no pain".
Weak prompt / Strong prompt
Weak prompt:
Give me 20 startup ideas.
This prompt produces a garbage dump of ideas without context and segments; All are solution-oriented, none are based on real pain.
Powerful prompt:
Your role: problem-focused customer discovery mentor. Segment:[small tour company owners]. Write the 8 most likely pains this group experiences while doing their job, each in the "segment-situation-pain-cost" pattern. Label each pain as a painkiller/vitamin and tell me how to verify the 3 strongest with a cheap one week test.
Approach
Nature of the output
Verifiability
start with solution
Cool but unfounded idea
low
Soliciting non-segmented opinions
General, not distinctive
low
Problem + segment + hypothesis
List of testable pain
high
Problem + cheap test design
Portable plan to the field
highest
Common mistakes
- Starting from the solution. Imagine the product and then look for problems; reverse and dangerous order.
- Doing it for "everyone". The wider the segment, the more blurred the pain becomes; Start small.
- Mistaking vitamins for painkillers. Falling in love with “it would be nice to have” ideas; Test payment willingness.
- Mistaking the AI list for real. AI's problem list is the hypothesis; No product is made without verification with real people.
- Not asking about frequency/cost. No one buys a solution for a problem that is rarely or freely endured.
Beware: AI may present an ordinary problem as a "big opportunity" in order to be sympathetic to you. Against confirmation bias, for every problem created, be sure to ask "where is the evidence, what are people spending now?" Ask the question yourself. If there is no evidence, the idea is still just a guess.
In summary
Startups die when they make products for the wrong problem; That's why it is necessary to start with the problem, not the solution. The good problem is frequent, painful and expensive. The painkiller/vitamin distinction is the most practical way to predict which pain will cause a willingness to pay. AI is a powerful aid in digging into a segment's day, translating pain points into testable hypotheses, and designing inexpensive tests. But the problem list that AI produces is a starting point; Real pain is only validated on the ground, by talking to people. Use AI as both producer and critic without falling in love with your own idea.
Application task
Choose a segment (as narrow as possible). Take the pain out of that segment from the AI with the “digging a day” template. Then, create at least 5 hypotheses with the "Turning a problem into a hypothesis" template and mark each one as a vitamin/painkiller. Derive a weekly test plan from the “Cheap test design” template for the 2 strongest analgesic hypotheses. Plan to take at least one of these tests with a real person within a week, and write down in advance which result will say "the harsh truth."
checklist
- [ ] Did I start with the problem of a particular segment and not with the solution?
- [ ] Have I turned every problem into a testable hypothesis?
- [ ] Have I filtered each hypothesis along the vitamin/painkiller axis?
- [ ] “What are people spending on this right now?” Have I asked the question to every hypothesis?
- [ ] Did I treat the AI list as a hypothesis to be tested in the field rather than reality?