Validating Your Business Idea with Punit Mehra

Most startup advice breaks in high-stakes markets.

“Move fast and iterate” only works when the cost of being wrong is low. In biotech, it isn’t. A single mistake can take years to surface, and cost tens of millions when it does.

In this week’s episode of the Predictable Revenue Podcast, host Collin Stewart spoke with Punit Mehra, co-founder of ALP AI, about what building in that kind of environment actually looks like.

ALP AI operates in a world where you can’t afford to learn by failing late. That constraint reshapes everything, from how you validate demand to how you price, sell, and build credibility.

This post breaks down those lessons and what they reveal about finding product-market fit when speed isn’t the advantage.

The Founder Trigger

Most startup stories begin with excitement: the big idea, the breakthrough, the moment everything clicks. That’s not what pushed Punit Mehra to start ALP AI.

When he describes the decision, he doesn’t mention inspiration, he does use the word “discomfort”. He could see exactly where his career was heading: a stable role, steady progression, increasing responsibility. On paper, it all made sense.

But as he put it, “I knew the trajectory… I knew where it was going. But I couldn’t change anything meaningful.” 

That clarity became the problem.

Because once the path is defined, it’s easy to follow it without questioning whether it’s the right one. Promotions come, compensation improves, and progress becomes predictable, but the scope for real impact often shrinks.

Staying meant committing to a path he hadn’t chosen, so he left without a fully formed vision or a guaranteed opportunity. A lot of founders start by deciding they’re no longer willing to stay where they are.

The Insight: A High-Stakes Problem

In most startups, failure is part of the process. Teams iterate quickly, and the cost of being wrong is manageable. Biotech operates under a different set of constraints.

Developing a single drug can take close to a decade and cost anywhere from $5 million to over $100 million. By the time something fails, years of research, capital, and momentum have already been committed.

For smaller companies, that failure is existential. Many are built around one or two core drug candidates, and if those fail, there’s often no way to recover. As Punit put it, “If they fail, they’re out.”

Even large companies aren’t immune. 

Pfizer reportedly spent around $750 million developing a drug that ultimately failed in late-stage trials. Beyond the financial loss, the opportunity cost is just as high, years of potential revenue that never materialize.

This changes how companies approach product development. In most industries, the goal is to improve over time. In biotech, the priority is avoiding catastrophic failure.

The Product Insight (Why ALP AI exists)

For Punit, the opportunity wasn’t just the science. His co-founder, Luca, had been deep in research, exploring how machine learning could be applied to protein design. The work was compelling, but what stood out was how often promising ideas failed once they moved beyond controlled environments.

Drugs that looked viable early on were still collapsing late on, after years of investment. By then, the cost of being wrong was already locked in.

That gap, between early promise and late-stage failure, is what turned the idea into a company.

Instead of building new drugs, they focused on a more immediate problem: identifying risk earlier, when there’s still time to act. The goal wasn’t to eliminate failure, but to change when it happens and what teams can do about it.

ALP AI sits upstream of the most expensive decisions, helping companies understand what might break before they fully commit. That can mean stopping early or adjusting before failure becomes irreversible.

Start with Demand Signals (Not Product)

One of the clearest signals that ALP AI was worth building came from the market.

At the time, the product was still early. The technology wasn’t fully built, and there was no polished offering. Instead of waiting, the team reached out to biotech and pharma companies through cold outreach.

The response was immediate.

Conversations started quickly, and instead of pushing back on an unfinished product, companies were asking when it would be ready and how they could use it. In a market known for being slow and difficult to access, that responsiveness stood out. It was urgent.

That’s what validated the direction. 

The problem already existed, buyers understood it, and it was important enough that they were actively looking for better solutions.

The takeaway is simple: demand doesn’t come from a finished product. It comes from solving a problem that already has attention and budget behind it.

Fit Into Existing Budgets

Early on, the team made a key decision: they didn’t try to invent a new pricing model. Instead, they started by understanding how their customers were already spending money.

Biotech companies were already paying to test for immune risk. There were existing methods, vendors, and, most importantly, an allocated budget. Even if those methods weren’t perfect, the line item existed.

That became the entry point.

Rather than forcing customers to create a new budget, ALP AI positioned itself within existing spend. If companies were already paying to assess risk, the question became whether there was a better way. 

From there, they added a second layer.

If their approach improved a drug’s chances of success, they shared in the upside through milestone-based pricing. That way, the economics reflected the same reality their customers faced: most of the value only exists if the drug succeeds.

Customers didn’t take on additional upfront risk, and ALP AI captured meaningful value only when it delivered impact.

Align With Customer Risk

The deeper constraints in biotech are spending and survival. Most companies are built around one or two drug candidates, if those succeed, the company can be worth hundreds of millions. 

If they fail, there’s often no fallback. That dynamic shapes how decisions are made and what good pricing looks like.

ALP AI didn’t position itself as a typical vendor charging upfront regardless of outcome. Instead, they aligned with the same risk their customers were taking on. There’s still an initial engagement, but a meaningful portion of value is tied to what happens next. If a drug progresses, they share in the upside. If it doesn’t, the economics are limited.

The best pricing models don’t just reflect value, they reflect customer risk. In markets like biotech, where outcomes are binary, alignment matters more than optimization.

Credibility > Speed

One of the biggest adjustments for the team was operational.

Most startup advice emphasizes speed: ship early, iterate quickly, and learn by doing. That works in software, where failure is cheap and reversible. In biotech, it doesn’t.

Every claim needs evidence, every result must be reproducible, mistakes compound over years, not weeks. As Punit put it, “credibility is the key currency.”

That changes how you build.

Instead of optimizing for speed, the team prioritized trust, moving deliberately, investing in data, and producing results that could withstand scrutiny from highly technical buyers. In this environment, credibility is the foundation.

The Sales Motion (Underrated but strong)

One part of ALP AI’s approach that often gets overlooked is how they’ve handled sales. On the surface, the tactics are familiar: cold outreach, conversations, and relationship-building. What’s different is the emphasis on timing.

In biotech, companies only become viable customers at specific points in their development cycle. A drug needs to be at the right stage, the risk needs to be relevant, and the budget must already be allocated. Outside that window, even a strong solution is easy to ignore.

That reality shifts the role of sales. Instead of pushing a product, the team focuses on understanding where each company is in its process and whether the problem is urgent. The goal is to be present when the timing aligns, not to force a decision before it does.

This pattern shows up across the business. 

The advantage doesn’t come from moving faster, but from aligning with how the market actually works, its timelines, its risks, and its constraints.

In that kind of environment, speed doesn’t create leverage, accuracy does. Not all startups benefit from moving fast. In markets where the cost of being wrong is high, some companies win by being right when it matters.

NO TIME TO READ?