A few years ago, building a software product required a serious commitment. You either spent months developing an MVP yourself or raised enough money to hire a team. Development was expensive, which naturally forced founders to think carefully about which ideas were worth pursuing. AI has changed that equation. Today, you can build a functional product in a weekend. Sometimes in a few hours. And that's incredible. But there's an unintended consequence: building has become so easy that we're increasingly skipping the question of whether anyone actually needs what we're building. The bottleneck has shifted from Can I build this? to Should I build this at all?
This fall, I joined AltaLab, a startup accelerator created by venture investor Igor Ryabenkiy and the AltaIR Capital team, with my startup, TheBold.vc. AltaIR's portfolio includes more than ten unicorns, including Miro, Deel, PandaDoc, and Turing. During the program, we worked through several frameworks designed to help founders evaluate startup ideas before committing significant resources to them.
With the team's permission, I'm sharing a few of those frameworks, along with my own observations about where they work, where they fall short, and how I'd apply them in the age of vibe coding.
1. Before You Build Anything, Ask Yourself One Question
One of our first assignments at AltaLab was surprisingly simple.
We were asked to list our startup ideas and evaluate each one using three criteria:
| Criterion | Question | Score |
|---|---|---|
| Passion | How badly do you want to solve this problem? | 1–5 |
| Expertise | Do you have the skills or a meaningful advantage in this space? | 1–5 |
| 10-Year Test | Are you willing to work on this problem for the next ten years? | Yes/No |
The first two scores are added together, for a maximum of ten points.
But there's a catch.
If an idea fails the 10-Year Test, the recommendation is to eliminate it, regardless of how highly it scores on everything else.
You might have deep expertise in a market and genuinely enjoy working on the problem. But if you can't imagine spending the next decade solving it, Igor recommends reconsidering the idea.
At first, this might sound excessive.
Especially today, when someone can build a small SaaS product in two weeks, find a few customers, and start generating revenue.
Why would you need to commit ten years to something you can build in a weekend?
The answer depends on what kind of business you're trying to create.
Not every good business needs to become a unicorn
AltaLab is primarily designed for founders who want to build scalable, venture-backed technology companies.
And for that kind of business, the 10-Year Test makes sense.
Building a venture-scale company usually requires years of persistence, uncertainty, and difficult decisions. You need a problem you care enough about to keep solving even when things go wrong.
I came into the program with an existing startup, TheBold. It's a platform that helps early-stage founders understand what specific investors are looking for, work toward their requirements, and eventually access investor meetings based on demonstrated progress. I've spent years on both sides of the venture ecosystem. I've raised several rounds of funding for my own projects. Over the years, I've reviewed more than a thousand startup pitches and made dozens of warm introductions to investors. So the problem I'm solving didn't appear because AI suddenly made it possible to build another application. It's a problem I've experienced and observed for years. That's why answering yes to the 10-Year Test is relatively easy for me.
Where I disagree with the framework
Imagine you discover a narrow B2B problem. There are a few dozen companies willing to pay $500 per month for a solution.
You could build the product in a couple of weeks, acquire customers, and create a small, profitable business.
Would that be a bad idea simply because you don't want to spend the next ten years working on it?
Of course not. It might be an excellent bootstrapped SaaS business, even if it would never make sense as a venture-backed startup.
That's why I'd add one question before applying the framework:
What kind of company are you actually trying to build?
- A small, profitable business?
- Or a venture-backed startup with the potential to grow 10x, 100x, or more?
Those are different strategies. They require different criteria for evaluating ideas. And if you're building a small product with AI, you probably shouldn't judge it using the same framework a venture capitalist uses to evaluate billion-dollar opportunities.
But if your ambition is to build a large company, the 10-Year Test can be a useful way to distinguish a temporary fascination from a problem you're genuinely committed to solving.
2. Your Excitement About an Idea Is Not Evidence of Demand
Once you've found a problem you're interested in solving, the next step is to find out whether anyone else cares. This is where another AltaLab exercise comes in: the Validation Check. The idea is to interview at least five potential customers and evaluate each conversation against five criteria.
| Criterion | Maximum score |
|---|---|
| The person genuinely experiences the problem | 1 |
| Their current solution is inadequate | 1 |
| They understood your proposed solution | 1 |
| They found the idea relevant or interesting | 1 |
| They expressed willingness to pay | 3 |
| Total | 7 |
Notice the weighting. Willingness to pay accounts for almost 43% of the total score. That's intentional. For a business, it's not enough for people to like an idea. They need to value the solution enough to exchange money for it. But there's a problem.
What people say they're willing to pay and what they actually pay are two very different things.
I'll come back to that in a moment.
How AltaLab interprets the results
After conducting the interviews, you calculate the average score. The framework suggests the following interpretation:
| Average score | Confidence level | Suggested action |
|---|---|---|
| 6–7 | High | Strong signal; consider committing significant resources |
| 5–5.9 | Medium | Run additional experiments over the next 60–90 days |
| Below 5 | Low | Reconsider the idea or explore another direction |
It's a useful structure, especially for founders who might otherwise rely entirely on intuition. But I think there's a danger in treating these scores as objective measurements of startup potential. Five people liking your idea doesn't mean you've found a market. And even if your average score is 6.5 out of 7, that doesn't guarantee any of those people will become paying customers. The quality of the interviews matters at least as much as the score. Which brings me to one of my favorite books about customer discovery.
3. Stop Asking People Whether They Like Your Idea
If you've worked in product management, you've probably heard of The Mom Test by Rob Fitzpatrick.
Its central argument is simple:
People often give you misleading answers, not because they're trying to deceive you, but because you're asking the wrong questions.
Imagine telling your mother about an app you want to build.
She loves you. She wants to support you. So she says, "That sounds like a great idea! I'm sure people would use it."
You leave the conversation feeling encouraged. But what have you actually learned about market demand? Almost nothing. The same thing happens with friends, colleagues, and even strangers. Most people don't want to disappoint you. And people are notoriously bad at predicting their own future behavior. They might genuinely believe they'll use your product, only to forget about it a week later. Or confidently say they'd pay for it, then change their minds when they see the checkout page.
The Mom Test offers three useful principles:
- Talk about the customer's life, not your idea.
- Ask about specific past experiences, not hypothetical future behavior.
- Listen more than you talk.
Let me show you how this works using TheBold.
The wrong question
"Would you use a platform that helps you find investors?"
Most founders planning to raise capital would probably say yes. But what have we learned? Only that finding investors sounds appealing. We still don't know whether the person has an urgent problem, how they're solving it today, or whether they're willing to pay for a better solution.
A better question
"Tell me about the last time you tried to find investors. How did you approach it?"
Now we're talking about actual behavior. From there, we can ask:
- How did you find investor contacts?
- How much time did you spend researching them?
- How did you determine which funds were a good fit?
- How many messages did you send?
- How many responses did you get?
- What tools did you try?
- Did you pay for any of them?
Suddenly, we're no longer collecting opinions about our hypothetical product. We're learning how people actually solve the problem today.
Imagine a founder tells you:
"I spent the last two weeks researching investors, built a spreadsheet with 200 contacts, sent 70 messages, and got three replies."
Now that's useful information.
We can investigate which parts of the process consume the most time, what they've already tried to automate, and what solutions they've paid for. But suppose the founder says:
"Actually, I haven't really started looking for investors yet."
That's a different situation.
For this particular research question, they're probably not the right interview participant.
Building a product for people who already experience a painful problem is very different from building something that first needs to convince people they have a problem. If you struggle to find people who are actively trying to solve the problem today, that should be a warning sign. The Mom Test doesn't contradict AltaLab's Validation Check. It complements it by helping you ask better questions and interpret the answers more carefully.
4. Willingness to Pay Is More Complicated Than a Yes or No
Let's return to the most heavily weighted criterion in AltaLab's framework: willingness to pay.
During my own customer interviews, I noticed something interesting. Different founders had very different expectations about how they should pay for a product that helps them access investors. Some considered $25–50 per month a comfortable subscription price.
Others mentioned $49–100 per month, provided the product could demonstrate meaningful value. But several founders suggested completely different pricing models. One said they would consider paying $100–200 for a relevant investor connection.
Another suggested $50–100 for a qualified investor meeting. At least ten respondents independently proposed some form of outcome-based pricing, including success fees. Initially, it would be easy to interpret these responses as evidence that founders prefer paying for results rather than subscriptions. But I think there's another possible explanation.
Maybe users aren't against subscriptions.
Maybe they're against paying before they have evidence that the product will deliver value.
Those are two very different problems. And they require different solutions. In one case, you might need to change your pricing model.
In the other, you might need to improve how you demonstrate value before asking for payment. That's why I wouldn't treat someone saying "I'd pay $50 per month" as strong validation on its own. It's a useful signal, but still only a hypothesis. Real behavior matters more.
Have they already paid for an alternative?
Have they tried to solve the problem manually?
Are they spending time or money on it today?
Will they make an actual commitment, rather than simply express interest?
These questions tell you much more about potential demand than asking whether your pricing sounds reasonable.
5. The Most Important Step: Decide What Would Make You Stop
There's one final principle I'd add to this process. Before you start building, decide what evidence would convince you that your idea is worth pursuing. And just as importantly, decide what evidence would convince you to abandon it. Here's a practical sequence I'd use:
Step 1. Evaluate your founder–problem fit.
Do you understand the market? Do you have relevant expertise or an advantage? Are you genuinely interested in solving this problem?
And if you're building a venture-backed company, would you be willing to spend years working on it?
Step 2. Define the problem in one sentence.
Not your product.
Not its features.
The actual problem experienced by a specific group of people.
Step 3. Find people who already experience that problem.
Ask when they last encountered it.
How did they solve it?
What was frustrating?
How much time or money did it cost them?
What alternatives have they tried?
What have they already paid for?
Step 4. Define your validation criteria in advance.
What would count as a strong enough signal to continue?
And what would make you reconsider the idea?
This last step is particularly important.
Because founders are remarkably good at explaining away negative results.
Maybe we interviewed the wrong people.
Maybe the prototype wasn't polished enough.
Maybe the onboarding was confusing.
Sometimes those explanations are correct.
Sometimes you really do need to change the onboarding, reposition the product, or experiment with a different business model.
But sometimes the explanation is much simpler:
You're building something people don't need badly enough.
And the earlier you discover that, the more time you have to find an idea worth pursuing.
Final Thoughts
AI has dramatically reduced the cost of building software. That's an extraordinary opportunity, especially for solo founders and small teams.
But it also makes it easier than ever to spend time creating products without sufficient evidence of demand. The ability to build something is no longer a compelling reason to build it. And a working MVP isn't the same thing as a validated business. The frameworks we explored at AltaLab are useful precisely because they force founders to examine the problem before falling in love with the solution.
Y
ou don't need to follow every framework literally. The right validation process will depend on whether you're building a venture-backed startup, a bootstrapped SaaS, or a small side project.
But the underlying principle remains the same:
Before you spend a weekend vibe coding your next idea, spend some time finding out whether anyone actually needs it.
***
Thanks to the AltaLab team for allowing me to share selected frameworks from the program. AltaLab is a free accelerator for early-stage founders. Igor Ryabenkiy has also published a free startup workbook and the book Unicorn Focus for founders who want to explore these ideas further.
