Posts


# Why Your Vibecoded Product Has a 99% Failure Chance (And Our Free 10-Minute Test That Checks For You)

> **Vibecoded product failure** happens not because the code is bad, but because vibecoding removed the natural friction that forced validation. Building is now so fast that teams skip demand testing, assumption checking, and risk analysis — shipping products nobody wants at record speed. The 90%+ startup failure rate didn't change; vibecoding just accelerated the path to failure.

Vibecoding is incredible. You describe what you want, and AI builds it. Features that took weeks now take hours. Full apps materialize from conversations.

But here's what nobody talks about:

**Vibecoding didn't change the 90%+ startup failure rate. It accelerated it.**

You're not building wrong. You're building wrong things, faster.

## The Speed Trap

In 2024, vibecoding went mainstream. Cursor, Claude, v0, Bolt—suddenly anyone could build a functional app in a weekend.

Indie Hackers exploded with "I built this in 48 hours" posts. Twitter filled with demos. ProductHunt launched 3x more products.

But something else happened: the failure rate didn't budge.

**CB Insights data says:**
- 42% of startups fail because "no market need"
- 29% fail because "ran out of cash" (often from building the wrong thing)
- 19% fail because "got outcompeted"

Notice what's missing? "Couldn't build the product."

Building was never the bottleneck. **Knowing what to build was.**

Vibecoding solved a problem that wasn't the problem.

## Why Vibecoded Products Fail

Let me describe the typical vibecoding failure arc:

### Week 1: The Idea
You have an idea. It feels exciting. You start prompting.

### Week 2: The Build
Features accumulate. Landing page goes up. Auth, payments, dashboards—all shipped.

### Week 3: The Launch
ProductHunt. Twitter. Reddit. Traffic spikes. Some signups.

### Week 4: The Silence
Usage drops. Churns begin. Nobody's coming back. Nobody's paying.

### Week 5-8: The Pivot
Maybe add this feature? Different positioning? Lower price?

### Week 12: The Graveyard
Project abandoned. "Wasn't the right time." "Market wasn't ready."

**This pattern repeats thousands of times per month.**

The builds are fine. The product/market fit was never there.

## The Three Killers of Vibecoded Products

### Killer #1: Unvalidated Assumptions

Every product is a stack of assumptions:
- "Users have this problem"
- "Users will pay to solve it"
- "Users will choose my solution"
- "My channel will reach users"
- "Users will retain"

Most vibecoded products have 20+ assumptions. Most founders can't list 5.

**The failure mode:** You build on top of untested assumptions. When they're wrong (and most are), the whole thing collapses.

### Killer #2: No Product Context

AI coding agents build what you tell them. They don't ask:
- "Is this feature validated?"
- "Does your target user actually want this?"
- "Does this conflict with your business model?"

You're the [product manager](/). If you don't provide context, the AI builds in a vacuum.

**The failure mode:** You build features that seem good in isolation but don't fit your product, users, or business.

### Killer #3: Speed-Induced Blindness

When building is fast, reflection is slow.

Traditional development had natural pause points—waiting for deploys, code reviews, standup meetings. These pauses forced you to think.

Vibecoding removes the pauses. You can act on every impulse immediately. Feature idea? Shipped in an hour. Pivot? Done by dinner.

**The failure mode:** You never stop to ask "should we build this?" because you're too busy asking "can we build this?"

## The 10-Minute Test

Here's a quick test to check if your vibecoded product is headed for the graveyard.

**Answer these 10 questions honestly (1 point each for "yes"):**

### Demand Questions
1. Have you talked to 5+ potential users about their problem (not your solution)?
2. Can you name 3 specific people who would pay for this today?
3. Do you know how your target users solve this problem currently?

### Solution Questions
4. Can you describe your core value prop in one sentence?
5. Have you written down your top 5 assumptions?
6. Do you know which assumptions are validated vs. guessed?

### Business Questions
7. Have you tested your price point with real users?
8. Do you know your target CAC and LTV?
9. Can you describe your distribution channel (not "launch on ProductHunt")?

### Competition Questions
10. Can you name your top 3 competitors and why users would choose you?

---

**Score yourself:**

| Score | Status |
|-------|--------|
| 8-10 | You've done the work. Build with confidence. |
| 5-7 | Significant gaps. Consider validating before building more. |
| 2-4 | High risk. Your assumptions are mostly untested. |
| 0-1 | You're gambling. Stop building and start validating. |

---

Most vibecoded products score 2-4. That's why most fail.

## What Changes Your Odds

### Level 1: Manual Validation

Before building, spend 48 hours on validation:
- Talk to 10 potential users
- Test pricing with a fake door
- Map your assumptions

This alone moves you from 10% success rate to maybe 25%.

### Level 2: Structured Risk Analysis

Run a [pre-mortem](/blog/posts/pre-mortem-template):
- List everything that could kill your product
- Rank by likelihood and impact
- Define kill criteria before you're emotionally invested

This catches risks your optimistic brain ignores.

### Level 3: Context-Driven Development

Give your AI coding agent [product context](/blog/posts/product-context-for-ai-agents):
- What's validated vs. assumed
- Who your users are (and aren't)
- What constraints matter

Context-aware agents don't just build fast. They build right.

## The Cutline Pre-Mortem (Free, 10 Minutes)

We built [Cutline](/pre-mortem) to solve this problem.

In 10 minutes, Cutline:

1. **Analyzes your product idea** - You describe what you're building
2. **Identifies failure modes** - AI finds what could kill it
3. **Ranks your assumptions** - By risk and uncertainty
4. **Generates experiments** - Tests to validate assumptions
5. **Creates personas** - Simulated users to interview
6. **Outputs a risk report** - Shareable with cofounders/investors

It's like having a skeptical co-founder who asks the hard questions you're avoiding.

### What Users Say

> "Cutline found three fatal flaws in my idea that I was too excited to see. Saved me months."
> — Indie hacker, killed a bad idea in 20 minutes

> "Ran the pre-mortem, realized my core assumption was untested. Validated it in a week, then built with confidence."
> — Solo founder, launched successfully

> "I was about to spend 6 weeks building. Cutline showed me the demand risk was critical. Ran a survey, found out nobody would pay. Pivoted before wasting time."
> — First-time founder, avoided a classic mistake

## The Alternative: Expensive Lessons

If you don't validate, you'll learn eventually. The lessons just cost more:

- **3 months building** → Find out nobody wants it
- **$5,000 in ads** → Find out nobody converts
- **6 months of runway** → Find out the market is too small
- **Your mental health** → Find out failure hurts

Or you spend 10 minutes on a pre-mortem and learn for free.

## Action Steps

**If you're about to vibecode something:**

1. Take the [10-minute test](#the-10-minute-test) above
2. If you score below 7, pause
3. Run a [free pre-mortem](/start) in 10 minutes
4. Get your risk report
5. Validate your riskiest assumption first
6. Then vibecode with confidence

**The code is easy. The validation is what matters.**

---

*Ready to test your idea? [Run a free pre-mortem](/start) in 10 minutes. It might save you months.*

---

## FAQ

**Q: But I've already built something. Is it too late?**

No. Run a pre-mortem on what you've built. Better to know now than after 6 more months.

**Q: What if the pre-mortem says my idea is bad?**

Then you saved yourself months of building the wrong thing. Kill it fast and move on to something better.

**Q: Does vibecoding itself cause failure?**

No. Vibecoding is a tool. The failure comes from skipping validation because building feels like progress. Vibecoding just makes it easier to skip.

**Q: How is this different from just doing customer interviews?**

Customer interviews are great but limited to who you can reach. A pre-mortem:
- Systematically covers all risk categories
- Generates risks you wouldn't think of
- Includes AI personas you can interview 24/7
- Creates a structured output you can act on

**Q: Is this really free?**

The pre-mortem is free. Full access to AI personas and deeper analysis requires premium, but the core risk assessment is free forever.


Why Your Vibecoded Product Has a 99% Failure Chance (And Our Free 10-Minute Test That Checks For You)


# How to Validate Your MVP in 48 Hours (Without Writing Code)

> **MVP validation without code** is the process of testing demand, pricing, and positioning for a product idea before writing a single line — using fake door landing pages, pre-mortems, direct outreach, and pricing experiments. A 48-hour validation sprint costs $50-100 and produces demand data, a risk analysis, user interviews, and a clear go/no-go decision, potentially saving 4-8 weeks of building the wrong thing.

The traditional MVP playbook: spend 4-8 weeks building something minimal, launch it, see if anyone cares.

The problem? You just wasted 4-8 weeks if no one cares.

**Better approach:** Validate demand, pricing, and positioning in 48 hours. Then decide if the MVP is worth building.

Here's the playbook.

## The 48-Hour Validation Sprint

### Hour 0-2: Define Your Hypothesis

Before testing anything, get specific about what you're testing.

**Fill in these blanks:**

1. **Problem statement:** "[Target user] struggles with [specific problem] because [root cause]."

2. **Solution hypothesis:** "[Product name] helps [target user] [achieve outcome] by [mechanism]."

3. **Willingness to pay:** "[Target user] will pay [$X/month] because [value justification]."

**Example:**
- Problem: "Indie hackers struggle with building products nobody wants because they skip validation."
- Solution: "Cutline helps indie hackers identify product risks before building by running AI-powered pre-mortems."
- WTP: "Indie hackers will pay $10/month because it's cheaper than wasting 3 months building the wrong thing."

**Deliverable:** One-page hypothesis doc

---

### Hour 2-8: Build Your "Fake Door"

A fake door is a landing page that sells a product you haven't built yet. It tests whether people *want* what you're offering.

**What to include:**

1. **Headline** - Clear value proposition (not clever, clear)
2. **Subhead** - Who it's for and what problem it solves
3. **3-5 benefits** - What users get (outcomes, not features)
4. **Social proof** - Even "Built by ex-Google PM" counts
5. **CTA** - "Join Waitlist" or "Get Early Access"
6. **Pricing hint** - Optional but useful for testing WTP

**Tools to use:**
- Carrd ($9) - Fastest for simple pages
- Framer (free tier) - Better design options
- Typedream (free tier) - Good templates
- Notion + Super.so - If you like Notion

**Deliverable:** Live landing page with email capture

---

### Hour 8-12: Run a Pre-Mortem

Before driving traffic, stress-test your idea. What could go wrong?

Use the [pre-mortem framework](/blog/posts/pre-mortem-template) to identify:

- **Demand risks:** Do people actually have this problem?
- **Product risks:** Will your solution actually work?
- **Channel risks:** Can you reach your target users?
- **Business model risks:** Does the math work?

**Why this matters now:** Your fake door test will tell you *if* people click. The pre-mortem tells you *why they might not buy*.

You can run a pre-mortem in 30 minutes manually, or use [Cutline](/pre-mortem) to run one in 10 minutes with AI assistance.

**Deliverable:** Ranked list of top 3 risks and kill criteria

---

### Hour 12-20: Drive Traffic

Now you need eyeballs on your landing page. Here's how to get 200-500 targeted visitors in 8 hours.

**Channel 1: Communities (4 hours)**

Post in communities where your target users hang out:
- Reddit (find relevant subreddits)
- Indie Hackers
- Twitter/X (with relevant hashtags)
- Discord servers
- Slack communities
- Facebook groups

**Template post:**
```
Hey [community] 👋

I'm building [product] for [target user].

Quick question: [genuine question about the problem]

I put together a landing page if anyone wants to see what I'm 
thinking: [link]

Would love feedback on whether this resonates with your experience.
```

**Don't spam.** Engage genuinely. Ask questions. Add value.

**Channel 2: Direct Outreach (4 hours)**

Reach out directly to 50-100 potential users:
- LinkedIn messages
- Twitter DMs
- Cold emails
- Friends of friends

**Template message:**
```
Hey [name],

Saw your [post/product/comment] about [topic]. I'm working on 
something that might help with [specific problem].

Would you be open to checking out a 2-minute landing page and 
telling me if this would be useful for you?

[link]

No pressure either way – just trying to validate if this is 
worth building.
```

**Deliverable:** 200-500 page visits

---

### Hour 20-24: Analyze Results

Now measure what happened.

**Metrics to track:**

| Metric | Target | Your Result |
|--------|--------|-------------|
| Landing page visits | 200+ | |
| Email signups | 10%+ conversion | |
| Replies to outreach | 20%+ response rate | |
| Positive sentiment | 50%+ of replies | |

**Signals to look for:**

✅ **Green flags:**
- Signup rate > 10%
- People asking "when does this launch?"
- People sharing the page
- Detailed questions about pricing/features

🚩 **Red flags:**
- Signup rate &lt; 5%
- "Cool idea" with no follow-up
- Questions like "why would I need this?"
- Silence from direct outreach

**Deliverable:** Metrics dashboard and signal analysis

---

### Hour 24-36: Run Pricing Tests

If demand signals are positive, test willingness to pay.

**Method 1: Email Survey**

Send to your waitlist signups:
```
Thanks for signing up for [product]!

Quick question to help us prioritize features:

If [product] could [main benefit], what would you expect to pay?

a) $0 - I'd only use this if free
b) $5-10/month
c) $10-25/month
d) $25-50/month
e) $50+/month

Just reply with a letter!
```

**Method 2: Van Westendorp Price Sensitivity**

Ask 4 questions:
1. At what price would this be too cheap (low quality)?
2. At what price is this a bargain?
3. At what price is this getting expensive?
4. At what price is this too expensive?

The intersection of answers reveals optimal pricing.

**Method 3: Fake Checkout**

Add a "Buy Now" button that goes to a "Coming Soon" page. Track click-through by price point. (Ethical note: Be transparent that you're validating.)

**Deliverable:** Price sensitivity analysis

---

### Hour 36-44: Get Qualitative Feedback

Numbers tell you *what*. Conversations tell you *why*.

**Run 5-10 user interviews:**

Book 15-minute calls with:
- People who signed up (why?)
- People who visited but didn't sign up (why not?)
- People from your outreach (their real reaction)

**Questions to ask:**
1. "What problem were you hoping this would solve?"
2. "How are you solving this problem today?"
3. "What would make this a no-brainer purchase?"
4. "What concerns would prevent you from buying?"

**Deliverable:** Interview notes with key themes

---

### Hour 44-48: Make the Call

Now decide: Build or pivot?

**Decision framework:**

| Signal | Build | Pivot | Kill |
|--------|-------|-------|------|
| Signup rate | >10% | 5-10% | <5% |
| Pricing validation | Will pay target price | Will pay, but lower | Won't pay |
| Interview sentiment | Excited, asking for launch | Interested, but concerns | Polite, not engaged |
| Pre-mortem risks | Manageable | Addressable with pivot | Fundamental |

**Your options:**

1. **BUILD** - Signals are strong, start the MVP
2. **PIVOT** - Interest exists but for adjacent problem/audience
3. **ITERATE** - Run another 48-hour sprint with adjusted positioning
4. **KILL** - Fundamental lack of demand, move on

---

## What You'll Have After 48 Hours

- ✅ Live landing page (reusable for launch)
- ✅ Email waitlist (early adopters)
- ✅ Demand data (signup rate, pricing)
- ✅ Risk analysis (pre-mortem output)
- ✅ User interviews (qualitative insights)
- ✅ Clear go/no-go decision

**Total cost:** Maybe $50-100 for tools and ads
**Total time:** 48 hours
**Potential savings:** 4-8 weeks of building the wrong thing

---

## Tools for Your 48-Hour Sprint

| Purpose | Tool | Cost |
|---------|------|------|
| Landing page | Carrd, Framer, Typedream | Free-$9 |
| Email capture | Buttondown, ConvertKit | Free |
| Pre-mortem | [Cutline](/pre-mortem) | Free |
| User interviews | Cal.com + Zoom | Free |
| Analytics | Plausible, Fathom | Free trial |
| Price testing | Typeform, Tally | Free |

---

## The Meta Point

Building an MVP without validation is gambling with your time.

In 48 hours, you can know:
- Does anyone want this?
- Will they pay for it?
- What could kill it?

Then you can build with confidence—or save weeks by killing a bad idea fast.

**The best founders don't just build fast. They validate faster.**

---

## FAQ

**Q: How do you validate an MVP without writing code?**

Use a 48-hour sprint: define your hypothesis, build a fake door landing page, run a pre-mortem for risk analysis, drive 200-500 targeted visitors, then analyze demand signals and run pricing tests. Total cost: $50-100.

**Q: What is a fake door test?**

A fake door is a landing page that sells a product you haven't built yet. It tests whether people want what you're offering. A signup rate above 10% is a green flag; below 5% is a red flag.

**Q: What signals tell you an MVP idea is worth building?**

Green flags: signup rate above 10%, people asking "when does this launch?", sharing the page, detailed pricing questions. Red flags: below 5% signup, "cool idea" with no follow-up, "why would I need this?", silence from outreach.

**Q: How much does MVP validation cost?**

$50-100 total for tools and ads across 48 hours. You get a live landing page, email waitlist, demand data, risk analysis, user interviews, and a go/no-go decision.

---

*Want to run the pre-mortem portion in 10 minutes instead of 30? [Try Cutline's AI pre-mortem—it's free](/start).*


How to Validate Your MVP in 48 Hours (Without Writing Code)


# Pre-Mortem Template: The 30-Minute Product Risk Assessment

> A **pre-mortem template** is a structured 30-minute exercise where you imagine your product has failed catastrophically and work backward through five risk categories — demand, product, channel, business model, and competitive — to identify why. Each category produces scored risks and testable assumptions. The output: your top 3 kill-or-fix risks, explicit kill criteria, and a go/no-go decision backed by structured analysis rather than optimism.

A post-mortem asks: "Why did this fail?"

A **pre-mortem** asks: "Why *will* this fail?"—before you've built anything.

It's the single most effective technique for catching product risks early. Here's a complete template to run one in 30 minutes.

## Why Pre-Mortems Work

Traditional planning is optimistic. You imagine success and work backward.

Pre-mortems are deliberately pessimistic. You imagine failure and ask what caused it.

This mental shift surfaces risks your optimistic brain suppresses:
- The assumption you're not questioning
- The competitor you're underestimating
- The user behavior you're wishfully assuming
- The technical debt you're ignoring

Research by psychologist Gary Klein found that pre-mortems increase the ability to identify reasons for future outcomes by **30%**.

## The 30-Minute Pre-Mortem Template

### Setup (2 minutes)

**Prompt:** "Imagine it's 12 months from now. Your product has failed completely. It's dead. Users left. Revenue is zero. The project is shut down."

Write down: *"[Product name] failed because..."*

Now you'll fill in the "because."

---

### Part 1: Demand Risks (8 minutes)

These are risks that no one wants what you're building.

**Questions to answer:**

1. **What problem are you solving?**
   - Write it in one sentence
   - Is this a vitamin (nice to have) or painkiller (must have)?

2. **Who has this problem?**
   - Be specific: not "businesses" but "B2B SaaS founders with 10-50 employees"
   - How many of them exist?

3. **How are they solving it today?**
   - If "nothing," that's a red flag (maybe it's not a real problem)
   - If "competitor X," why would they switch?

4. **What's your evidence of demand?**
   - Interviews conducted: ___
   - Waitlist signups: ___
   - LOIs or pre-orders: ___
   - If blank, this is your #1 risk

**Template output:**

| Risk | Likelihood (1-5) | Impact (1-5) | Evidence |
|------|------------------|--------------|----------|
| Target users don't have this problem | | | |
| Problem isn't painful enough to pay for | | | |
| Existing solutions are good enough | | | |
| Market is too small | | | |

---

### Part 2: Product Risks (8 minutes)

These are risks that your solution doesn't actually solve the problem.

**Questions to answer:**

1. **What's your core value proposition?**
   - In one sentence, why is your solution better?
   - Is this a 10x improvement or incremental?

2. **What must be true for your product to work?**
   - List 3-5 critical assumptions
   - Which have you validated?

3. **What's the simplest version that delivers value?**
   - Can you describe a 1-week MVP?
   - If not, you might be overbuilding

4. **What will users complain about?**
   - Every product has weaknesses
   - What are yours?

**Template output:**

| Assumption | Confidence (1-5) | Validated? | Test to Run |
|------------|------------------|------------|-------------|
| Users will understand the value prop | | | |
| Core feature works as expected | | | |
| Users will complete onboarding | | | |
| Users will return after first use | | | |

---

### Part 3: Channel Risks (6 minutes)

These are risks that you can't reach your target users.

**Questions to answer:**

1. **How will users find you?**
   - Primary channel: ___
   - Do you have experience with this channel?

2. **What's your expected CAC?**
   - If you don't know, estimate
   - Is this sustainable given your price point?

3. **What's your unfair advantage in distribution?**
   - Existing audience?
   - Partnerships?
   - SEO moat?
   - If "none," distribution is a major risk

**Template output:**

| Channel | Expected CAC | Confidence | Backup Plan |
|---------|--------------|------------|-------------|
| | | | |

---

### Part 4: Business Model Risks (6 minutes)

These are risks that the math doesn't work.

**Questions to answer:**

1. **What will you charge?**
   - Price point: ___
   - How did you decide this?

2. **What does it cost to serve one user?**
   - Infrastructure: ___
   - Support: ___
   - AI/API costs: ___

3. **What's your target LTV:CAC ratio?**
   - Below 3:1 is usually unsustainable
   - Have you modeled this?

4. **When do you break even?**
   - Runway: ___
   - Required users/revenue: ___

**Template output:**

| Metric | Assumption | Validated? | Kill Threshold |
|--------|------------|------------|----------------|
| Price point | | | |
| Conversion rate | | | |
| Churn rate | | | |
| CAC | | | |

---

### Part 5: Competitive Risks (5 minutes)

These are risks that someone else wins instead.

**Questions to answer:**

1. **Who are your direct competitors?**
   - List top 3
   - What do they do better than you?

2. **Who could enter this market?**
   - Big tech risk?
   - Well-funded startup risk?

3. **What's your moat?**
   - Network effects?
   - Data advantage?
   - Brand?
   - If "none," you're in a race

**Template output:**

| Competitor | Their Advantage | Your Advantage | Threat Level |
|------------|-----------------|----------------|--------------|
| | | | |

---

### Synthesis (5 minutes)

**Step 1: Rank your risks**

Review everything you wrote. Pick the **top 3 risks** that could kill your product.

1. _______________
2. _______________
3. _______________

**Step 2: Define kill criteria**

For each top risk, define what would make you stop:

| Risk | Kill Criterion | How to Test |
|------|----------------|-------------|
| Risk 1 | If ___, we stop | |
| Risk 2 | If ___, we pivot | |
| Risk 3 | If ___, we pause | |

**Step 3: Make a decision**

Based on your pre-mortem:

- [ ] **GO** - Risks are manageable, proceed with validation plan
- [ ] **CONDITIONAL GO** - Proceed only if [specific test] passes
- [ ] **PAUSE** - Need more information before deciding
- [ ] **NO GO** - Risks are too high, kill or pivot

---

## Pre-Mortem Output Example

Here's what a completed pre-mortem looks like:

**Product:** AI-powered expense tracking for freelancers

**Top 3 Risks:**
1. Freelancers don't track expenses (behavior change required)
2. Market is too small (&lt; 1M potential users)
3. Competitors offer expense tracking as a feature, not a product

**Kill Criteria:**
1. If &lt; 5% of surveyed freelancers say they'd pay, stop
2. If TAM &lt; $10M, pivot to adjacent market
3. If 3+ competitors launch similar features in 6 months, differentiate or stop

**Decision:** CONDITIONAL GO - Run demand survey first

---

## Running This With AI

You can run this template manually, or use [Cutline's AI-powered pre-mortem](/pre-mortem) to:

- **Auto-generate risks** based on your product description
- **Score assumptions** by likelihood and impact
- **Get persona feedback** from simulated target users
- **Generate experiment plans** for each risk
- **Export a shareable report** for stakeholders

The AI pre-mortem takes 10 minutes instead of 30, and catches risks you might miss.

---

## When to Run a Pre-Mortem

- **Before starting a new product** - Catch fatal flaws early
- **Before major features** - Validate assumptions before building
- **Before fundraising** - Know your risks before investors ask
- **Quarterly** - Revisit as your product evolves

---

## Download the Template

Want a copy of this template? The full pre-mortem framework is built into Cutline.

[Run a Free Pre-Mortem →](/start)

---

---

## FAQ

**Q: How long does a pre-mortem take?**

A manual pre-mortem takes 30 minutes: setup (2 min), demand risks (8 min), product risks (8 min), channel risks (6 min), business model and competitive risks (11 min), synthesis (5 min). AI-assisted takes about 10 minutes.

**Q: What are the five parts of a pre-mortem template?**

Demand risks (does anyone want this), product risks (does the solution work), channel risks (can you reach users), business model risks (does the math work), and competitive risks (will someone else win).

**Q: When should you run a pre-mortem?**

Before new products, before major features, before fundraising, and quarterly as your product evolves.

---

*Pre-mortems don't predict the future. They prepare you for it. The best founders know what could kill them—and build anyway, with eyes open.*
