Product Leadership & Predicting the Future

Every product leader is secretly a futurist.

Look at the actual job:

• Vision is defining a future that customers want to be a part of

• Strategy is probabilistic bets executed in a sequence to achieve your vision

• Financial planning is forecasting future revenue based on acquisition, expansion and retention

• Hiring is adding headcount to create velocity against demand you have not yet captured

This is the uncomfortable reality of leadership. You use the information you have, make the best decisions you can, and avoid analysis-paralysis at all costs. Speed matters. So does betting on the right vectors. A perfectly laid plan that arrives too late is still a bad plan.

Underneath the meetings and metrics, product leadership is a repeated act of taking positions on futures that have not happened yet.

We’re judged by the quality of those bets: how clearly we understand the odds, how quickly we learn, and whether our decisions create better outcomes for customers and the business.

Vision is not one skill

How do some of the most fabled visionaries make decisions? I find it helpful to study the patterns associated with some of the most prolific business minds in our domain: Steve Jobs, Jeff Bezos, Elon Musk, and Ray Kurzweil.

One thing these leaders seemed to understand well is what type of bet they were making. Here are four different approaches:

1. Experience bets. (Steve Jobs)

2. Durability bets. (Jeff Bezos)

3. Constraint bets. (Elon Musk)

4. Curve bets. (Ray Kurzweil)

These labels are shorthand. Jobs made more than experience bets, Bezos made more than durability bets, and so on. The value is in having different ways to reason about an uncertain future.

1. Experience bets: what will feel obvious once it exists?

Jobs was betting on an experience. Before the iPod, portable music already existed. MP3 players existed. Digital music existed. Storage was improving. Apple didn’t invent any of that. The insight was that the whole mess could be compressed into a human want: “1,000 songs in your pocket.”

That’s a forecast about behavior. Jobs was betting that if the experience became simple enough, people would stop thinking about music on shelves, discs, etc. They would think of it as something that they could carry with them wherever they went.

That kind of vision comes from taste.

Taste, however, can be a dangerous word to describe vision because it sounds like personal preference. In product, taste is earned pattern recognition. It comes from watching people struggle, noticing where they’re delighted, a deep understanding of your market, and knowing which technical capabilities can be rearranged into something that suddenly feels obvious.

An experience bet is strongest when your customer’s reaction isn’t “that’s impressive,” but “of course it should work this way.”

For example, our team at Twindo is making an experience bet for architects, engineers, and construction professionals who use iPhones and iPads to capture the messy physical world and turn those captures into 3D computer-aided design files.

Several enabling capabilities have started to converge. On supported devices, LiDAR now ships in the hardware. LiDAR emits infrared light, measures how that light returns, and helps the device understand depth. When you combine that depth data with video and positional tracking as someone moves through a space, you can reconstruct interior and exterior environments in 3D. Pair that with modern computer vision, and the output stops feeling like a rough 3D scan and starts feeling like a usable digital representation.

The customer doesn’t want “a scan.” They want a faithful way to revisit reality after they've left the site. They want to find all the electrical outlets in a room. They want to make design choices without driving back across town. They want to make progress while the physical world is no longer in front of them.

So the experience bet is not “LiDAR on an iPhone.” The experience bet is that virtual revisits can become simple, trustworthy, and high-fidelity enough that teams can make real decisions without always having to return to the job site. The emerging technologies make the value obvious: fewer return trips, faster decisions, and less wasted time.

2. Durability bets: what won’t change?

Bezos bet that some customer desires would remain stubbornly durable even as the internet changed everything around them.

People would want more selection. Lower prices. Faster delivery. Less friction. More trust.

That framing worked because it paired enormous uncertainty about the internet with unusual clarity about the customer. The technology was constantly evolving. The customer desire was constant.

This is a useful lesson because product teams are often distracted by what is changing. New tools. New competitors. New channels. New platform shifts. New urgency.

Great strategy often starts with what is not changing.

If your roadmap doesn’t focus on a durable customer need, it may just be metabolizing urgency.

I saw this clearly at Inspirato, a luxury travel company. The durable need was simple: customers wanted more of their vacation spent experiencing destinations with friends and family, and less of it spent managing their trip.

Before the trip, guests needed confidence that the important details were handled. During the trip, they needed their itinerary, directions, contact information, and charges to be clear without coordinating across emails, phone calls, paper documents, and concierge conversations. At departure, they wanted to settle up without turning the end of a vacation into an administrative task.

We used mobile technology to serve that durable need.

Groceries could be ordered through the app and stocked before guests arrived. Itineraries were laid out clearly in a calendar so the trip had shape without feeling over-managed. Directions and contact information were available when they needed them. The folio updated in real time and could be paid when they left.

The technology was mobile. The durable need was time.

More time enjoying the place. More time with the people they came with. More certainty that the trip was being handled.

A durability bet asks: what customer need will still matter ten years from now, and how can today’s technology serve it better?

3. Constraint bets: what’s possible when the bottleneck breaks?

Musk’s vision is often described as a bet on electric cars, reusable rockets, and cost curves. But the more useful lens is how he attacks the system around the product.

His first-principles question is simple: what is this thing actually made of, and why does it cost so much more than it should?

That question leads to a different approach to product strategy.

Sometimes the bottleneck is not customer demand or technical possibility. Sometimes it’s the length and complexity of the supply chain: too many integration dependencies, too many fees from too many vendors, too much distance between inputs and outputs.

Tesla and SpaceX didn’t integrate everything because “in-house” is morally superior. They integrated where speed, cost, quality, or learning were strategically constrained.

I saw this first-hand working in ad-tech, where ad transactions are facilitated through a complex chain of supply- and demand-side technologies. Each provider’s innovation was constrained by how it integrated with others in the supply chain. Each charged a fee for its part.

Every boundary between companies became a boundary around data, optimization, latency, and most importantly, experimentation. When no one owns the full system, everyone optimizes their slice of the transaction. The result was lots of local innovation but too much system-level drag. And as the market matured, it sought to squeeze these inefficiencies out of the system.

More and more of the dominant players pursued vertical integration: owning supply, demand, and everything in between. Integration was not easy or inherently more efficient to manage. The constraint was the operating model itself.

A constraint bet asks: what part of the system must we own because it determines how fast we can learn?

That is a powerful product question. It forces you to look beyond features and ask what controls the rate at which the whole system improves.

4. Curve bets: what are we reading linearly that is actually compounding?

Kurzweil gives us one more lens: humans often misread exponential change.

We look at the latest visible point and draw a straight line.

AI is the obvious example. A team tries a model, gets a poor result, and overgeneralizes from the snapshot. Another sees a magical demo and assumes transformation is immediate. Both are treating today’s capability as a stable fact instead of one point on a curve.

Some of the strongest engineers I have worked with have had negative experiences delegating to AI models. They get sloppy code that doesn’t meet their standards, then become skeptical or write it off entirely.

You can empathize with this.

Engineering is a discipline of precision. If the point-in-time capability yields a result you have to rework, you end up spending more time correcting than doing it yourself. And many developers have worked their entire careers perfecting their craft. So the prospect of AI taking some of this work off their hands means they have to reinvent themselves and find new ways to contribute value.

The trap is overgeneralizing from the current gap. A bad experience with today’s model may be real evidence about today’s workflow, but weak evidence about where the capability will be over the next few model cycles.

On the other hand, I have personally overestimated what AI can do at the frontier.

When I was experimenting with 3D Gaussian Splatting, a technique for creating photorealistic 3D scenes from captured imagery, I built an automated evaluation loop to improve a reconstruction of my home against a target metric. The loop produced only marginal gains and consumed a meaningful amount of GPU spend.

That experience was humbling.

AI is strongest where the solution space is rich with prior examples. At the edge of research, where progress depends on new insight rather than pattern recall, deep expertise is still a bottleneck.

That is the discipline of a curve bet.

You have to acknowledge the curve, name the bottleneck, and keep enough exposure to learn as both change.

Final thoughts

Product leadership is a repeated act of betting on the future.

The goal is to place better probabilistic bets, understand what kind of bet we’re making, and learn fast enough to adjust before the market does it for you.

That is why thinking in frameworks matters.

• Jobs was betting on an experience: what will feel obvious once it exists?

• Bezos was betting on durable customer behavior: what won't change?

• Musk was betting on constraints: which limits are real, and which ones are temporary?

• Kurzweil was betting on curves: where does progress compound faster than intuition expects?

Different futures require different logic.

A useful vision is not just a strong opinion about where the world is going. It names the mechanism that could make that future real. Better experience. Durable demand. A broken constraint. A compounding curve.

The real job is sequencing those bets, sizing them appropriately, and updating as reality responds. The customer doesn’t want your theory of the future. The customer wants the product that makes that future useful.

The mistake is confusing vision with certainty, or conviction with refusal to learn.

I believe the best product leaders hold a direction firmly and their assumptions lightly. They take positions on futures that haven’t happened yet, then refine those positions as new evidence arrives.

That is how vision becomes execution.

What frameworks are you using to place better bets on the future?