Get the permanent version delivered to your inbox.
Your handout is on the way.
Section 01
Your sprint team (no hiring required)
Five roles. Each defined by three qualities and one decision they own. Reference them by name in the prompts that follow. Your AI tool becomes each role when you tell it to.
You leave with: 5 named roles to assign your AI tool.
Skip if you already prompt your AI tool by persona name.
The team is not five people. It is five lenses on the same problem.
Section 02 / Day 1 of 5
Understand
I have seen strong validation done in 3 days without AI. The tools are better now.
2 to 3 hours. By the end of today you have a clear problem statement, a target customer profile, and a testable hypothesis.
You leave with: a problem statement, customer profile, and hypothesis.
Prompt: As the Strategy Lead
I want to validate a product idea before building it.
The problem I am trying to solve: [DESCRIBE THE PROBLEM IN 1-2 SENTENCES]
Help me:
1. Clarify this problem statement (make it specific and testable)
2. Identify who experiences this problem most acutely
3. Quantify the pain (how much time, money, or frustration does this cost?)
Be specific. Challenge my assumptions if they are vague.
Prompt: As the Customer Analyst
Based on the problem: [PASTE YOUR PROBLEM STATEMENT]
Research this market:
1. WHO has this problem? Demographics, job titles, behaviors, where they spend time online.
2. WHAT solutions exist today? Direct competitors, indirect competitors, DIY workarounds.
3. WHY haven't existing solutions worked? What is missing, too expensive, or too complicated?
I need this to write landing page copy later. Be thorough.
After Day 1 you should have: a one-sentence problem statement, a customer profile, a competitive landscape, and a hypothesis in the format "If I build [X], then [target] will [action] because [reason]."
Section 02 / Day 2 of 5
Debate
2 to 3 hours. Generate five solutions, stress-test them, pick one.
You leave with: one selected solution and three risks to watch.
Prompt: As the Builder
Problem: [PASTE PROBLEM STATEMENT]
Target Customer: [PASTE CUSTOMER PROFILE SUMMARY]
Gap in Market: [PASTE KEY INSIGHT]
Generate 5 different solution approaches. For each:
1. Name it (catchy, memorable)
2. One-sentence description
3. How it solves the problem differently
4. Rough effort to build (Low / Medium / High)
5. Potential revenue model
Include at least one unconventional approach.
Prompt: As the Devil's Advocate
Here are 5 solution approaches for [PROBLEM]:
[PASTE THE 5 SOLUTIONS]
For each solution:
1. Why might this fail?
2. What is the biggest risk?
3. What assumption is most likely to be wrong?
4. Who would NOT buy this and why?
Be harsh. I would rather know now than after building.
Prompt: As the Strategy Lead
Recommend ONE solution to test first. Criteria:
1. Fastest to validate (can test with a landing page)
2. Clearest value proposition (easy to explain)
3. Most defensible (hardest for competitors to copy)
4. Best fit for target customer
List your reasoning briefly. State any assumptions. If it is a close call, explain the trade-offs.
After Day 2 you should have: one selected solution, three risks to watch, and a testable hypothesis for Day 3.
Section 02 / Day 3 of 5
Draft
3 to 4 hours. Build your landing page and ad copy. The Builder leads today.
You leave with: a live landing page and ad copy ready to launch.
Prompt: As the Builder
Write landing page copy for:
Solution: [PASTE SELECTED SOLUTION]
Target Customer: [PASTE CUSTOMER PROFILE]
Key Value Prop: [ONE SENTENCE]
Include:
1. HEADLINE (8 words max, speaks to the pain)
2. SUBHEAD (explains the solution in one sentence)
3. 3 BENEFITS (not features: what do they get?)
4. SOCIAL PROOF placeholder (what type would work best?)
5. CTA BUTTON TEXT (action-oriented, specific)
6. OBJECTION HANDLER (address the number one reason they would hesitate)
Write 3 headline variations. You will A/B test later.
Prompt: Ad copy
Write ad copy to drive traffic to the landing page.
Target: [CUSTOMER PROFILE]
Solution: [SOLUTION]
Platform: [Facebook / LinkedIn / Google]
Focus on the PAIN, not the solution.
People search for problems, not products.
For Facebook/LinkedIn: 3 primary text variations (125 chars ideal) + 3 headlines (40 chars max).
For Google: 3 headlines (30 chars each) + 2 descriptions (90 chars each).
After Day 3 you should have: a live landing page and ad copy ready to launch.
Section 02 / Day 4 of 5
Test
1 hour setup, then wait. Launch ads and resist the urge to check every 10 minutes.
You leave with: ads running, analytics tracking, and a patience timer.
Prompt: Campaign setup
Help me set up an ad campaign:
Platform: [Facebook / LinkedIn]
Budget: [AMOUNT] total ([PER DAY] for [NUMBER] days)
Goal: Landing page traffic + conversion tracking
Target Audience: [PASTE CUSTOMER PROFILE]
Step-by-step setup guide. Include:
1. Best campaign objective for validation
2. Recommended audience size
3. Best times to run ads
4. How to structure ad sets
5. When to check results (not too early)
Do not do this while ads are running
Do not share the link with friends or family. Personal network equals biased data.
Do not change the landing page mid-test. One variable at a time.
Do not draw conclusions before 48 hours. Small samples lie.
Do not optimize for clicks. You want conversions, not traffic.
After Day 4 you should have: ads running, analytics tracking, and a 48-hour patience timer set.
Section 02 / Day 5 of 5
Learn
Sometimes the kill signal is clear by Day 4. That can save months of build time.
2 hours. The Signal Reader takes the lead. Read the data, make the call.
You leave with: a build, pivot, or kill decision backed by data.
Prompt: As the Signal Reader
Help me analyze the results:
AD PERFORMANCE:
- Impressions: [NUMBER]
- Clicks: [NUMBER]
- CTR: [PERCENTAGE]
- Cost per click: [AMOUNT]
LANDING PAGE:
- Unique visitors: [NUMBER]
- Avg time on page: [SECONDS]
- Bounce rate: [PERCENTAGE]
- CTA clicks: [NUMBER]
- Conversion rate: [PERCENTAGE]
QUALITATIVE (from Hotjar or session recordings):
- Common behaviors observed: [DESCRIBE]
- Where people dropped off: [DESCRIBE]
- Anything surprising: [DESCRIBE]
Based on this data:
1. Is there demand signal? (Yes / No / Maybe)
2. What worked?
3. What did not work?
4. What would you test next?
List your reasoning briefly. State any assumptions. Give a confidence rating from 1 to 10 only if the evidence is thin or mixed.
Decision framework
Signal
Meaning
Action
CTR > 2%, CTA clicks > 5%
Strong demand
Build the MVP
Mixed signals, one metric strong
Partial validation
Pivot one variable, retest
CTR < 1%, bounce > 80%
No demand
Kill this, take the lesson
After Day 5 you should have: a clear build/pivot/kill decision backed by data, not opinions.
The sprint gives you a decision. The weekly loop gives you momentum. What follows turns a one-time experiment into a continuous signal.
Section 03
Weekly agent prompts
Run these every Monday morning. 10 minutes. The agent reads your data so you can decide what to focus on this week.
You leave with: weekly talking points and hypotheses.
Skip if you already run weekly customer data synthesis.
Prompt 1: Weekly data synthesis (as the Signal Reader)
As the Signal Reader, you are my weekly data analyst. Review the following analytics data and surface what matters.
DATA:
[PASTE YOUR ANALYTICS EXPORT OR SCREENSHOT DESCRIPTION]
- Source: [PostHog / GA4 / Hotjar / Mixpanel]
- Period: last 7 days
- Key pages: [LIST YOUR MAIN PAGES]
Produce:
1. TOP 3 SIGNALS worth investigating this week, ranked by behavioral evidence (not volume)
2. For each signal: what happened, why it matters, and one hypothesis to test
3. ONE PATTERN across all three signals (if one exists)
Rules:
- Behavior over volume. 5 users completing a flow matters more than 500 bounces.
- Name specific pages, buttons, or flows. No vague "engagement improved."
- If the data is insufficient, say so. Do not fabricate insights.
Takes 10 minutes. Paste your analytics export or describe what you see. The agent does the pattern-matching.
Section 03 / continued
Interview talking points
Feed the synthesis output into this prompt. Get hypotheses and questions ready for customer conversations this week.
Prompt 2: Interview preparation (as the Customer Analyst)
Based on this weekly synthesis:
[PASTE OUTPUT FROM PROMPT 1]
Produce:
1. THREE HYPOTHESES to verify in conversation this week
Format: "We believe [user segment] struggles with [problem] because [evidence from data]. We will know we are right when [observable behavior]."
2. FIVE INDIRECT QUESTIONS that surface truth without leading
- No "Would you use X?" (they will say yes to be polite)
- No "Do you like Y?" (opinions are not behavior)
- Use: "Tell me about the last time you..." / "Walk me through how you..." / "What happened after..."
3. ONE THING TO WATCH FOR in body language or tone (hesitation, enthusiasm, confusion)
The best interview question you can ask is: "Tell me about the last time that happened."
Section 04
Interview framework
Inspired by Teresa Torres (Continuous Discovery Habits). One principle: never ask if they would use your feature. Watch what they already do.
Read once, reference during conversations. Skip if you do 5+ customer interviews per month.
You leave with: questions that surface truth without leading.
Questions test hypotheses. Hypotheses come from data. Data comes from the weekly agent.
Three question types
Type
Structure
Surfaces
Story
"Tell me about the last time you..."
Real behavior, not aspirations
Behavior
"Walk me through how you..."
Process, friction points, workarounds
Contrast
"What is different about the times when..."
Context that triggers the problem
Hypothesis format
We believe [user segment] struggles with [problem] because [evidence from agent].
We will know we are right when [observable behavior].
Section 04 / continued
Question bank
Reusable templates. Fill in the blanks. Never ask directly what you want to know.
Discovery questions
#
Question template
1
"Tell me about the last time you tried to [solve this problem]."
2
"What did you do right before you [took that action]?"
3
"Walk me through how you currently handle [this task]."
4
"What was the hardest part about [that experience]?"
5
"How did you decide between [option A] and [option B]?"
6
"What happened after you [completed that action]?"
7
"When was the last time this problem cost you time or money?"
8
"What would need to be true for you to switch from [current solution]?"
Never ask these
"Would you use X?" They will say yes. People are polite. Their calendar tells the truth.
"Do you like Y?" Liking is not buying. Ask what they did, not what they think.
"How much would you pay for Z?" They will lowball. Look at what they already pay for substitutes.
Section 05
Tools list
Curated for solo founders. Free tiers first. Self-hosted options for when you need full control.
Managed free tiers are listed first. They let you focus on validation instead of server maintenance. Move to self-hosted when you need full data ownership or cost control at scale.
Section 06
Further reading
You do not need all of these. Pick the one closest to where you are stuck.
Pick one. Read it this month. Skip if you have read 3+ of these.
44 experiment cards ranked by evidence strength. Extends Day 4 and 5.
Want this offline? Get the PDF.
One email. No spam.
Your handout is on the way.
The loop
This is not a one-time sprint. It is a weekly loop.
Run the sprint to validate. Then run the weekly agent prompts to keep learning. Every Monday: read the data, form hypotheses, talk to users all week. Repeat.
The tools change. The loop does not.
I help founders run validation sprints and set up AI agent workflows at byJed.com.
Review my calendar, open tasks, and last week's outcomes.
1. What is the ONE thing I should finish this week that moves the needle most?
2. What should I say no to or defer?
3. What am I avoiding that I should face?
List your reasoning briefly. Be direct.