Imagine having an idea for a new product, and today you can turn it into a prototype within hours, testing it without waiting for a tech team or allocating a massive development budget. What used to take weeks of coordination and execution is now achievable with significantly less effort.

This shift opens up vast opportunities for companies and product teams, but it also raises a question worth pondering: If building products has become easier, have we become better at choosing what is actually worth building?

This question struck me while watching the episode (Artificial Intelligence in Product Management) of the podcast (Zero to One), featuring Amin Fadhel, VP of Product at Salla, and Abdulrahman Hareth, in a discussion bridging technological advancement with the mindset and workflows inside product teams.

Among the ideas and examples presented, one point stands out as the ideal starting point: AI may shorten the path to building a product, but it does not guarantee that this product will find its users or deliver real value to them.

AI lowers the cost of building... but what about the product's value?

To understand the scale of this shift, one can look at what the internet did to the distribution cost of digital products. Companies were suddenly able to reach a massive audience without needing the physical infrastructure required by traditional distribution channels.

Today, AI is unlocking a similar transformation on another front: the cost of building products.

Instead of needing an entire team to execute every experiment, a small team can now leverage AI tools to produce prototypes and multiple iterations, reclaiming the time once spent on lengthy execution tasks.

However, a lower cost of building does not necessarily mean higher value in what we build.

In one example that caught my attention during the episode, Amin Fadhel brought up the idea of building an app similar to Uber or Careem using AI tools. Delivering the app in a short time might seem impressive, but what if you don’t have a plan to acquire customers or onboard service providers? And what would make a user choose your app over existing solutions?

Here, speed of building becomes less important than the questions that precede it: What problem does the product solve? What distinct value does it offer? And how will it reach the people who need it?

This example illustrates that having a technically functional product does not mean having a viable business. There are other aspects that cannot be shortcut by writing code, such as distribution, customer acquisition, building trust, and crafting an experience that keeps users coming back.

The problem is no longer always in our ability to build the solution, but in knowing whether the solution is worth building in the first place.

When does AI speed become a true advantage?

The value lies not just in building faster, but in learning faster.

This is a fundamental difference in product management approach. Instead of investing months in building a full initial version, AI can be used to test a specific concept and see if users are interested before committing larger resources to develop it.

Among the ideas mentioned in the episode was testing a model for a specialized school-run service, similar to ride-hailing app models. The idea here isn’t to build a new app simply because you can, but to use rapid execution to see if there is genuine demand for this kind of service.

This brings us back to the purpose of a Minimum Viable Product (MVP): not to prove that the team can build the product, but to test a hypothesis regarding the user and the problem.

AI can be utilized during this stage through:

• Building customer journey prototypes and testing them.

• Experimenting with multiple concepts for a new feature prior to development.

• Testing different messaging and experiences.

• Collecting, organizing, and extracting patterns from feedback.

• Shortening the time between posing a hypothesis and gathering initial indicators about it.

In this sense, speed becomes a means to reduce the cost of learning, rather than merely a way to inflate the number of releases.

Yet it can turn into a trap when the team gets caught up in producing successive prototypes and experiments without knowing what they are trying to prove. At that point, the work looks busier, but the team isn’t necessarily moving closer to solving a real problem.

How does AI change the day-to-day tasks of a product manager?

One of the most practical aspects discussed by the guests was the impact of AI on the time a product manager spends on daily responsibilities.

In the past, transforming an idea into a testable experiment required writing specifications and requirements, coordinating with engineering, phasing the scope, and waiting for the initial version to be built.

Today, AI can help streamline parts of this process.

Key use cases include:

1. Research and discovery

AI tools can be used to access preliminary market insights, research comparable solutions, explore competitor experiences, and identify trends worth examining.

2. Writing product requirements

Ideas and notes can be converted into more structured documentation, drafting use-case scenarios and edge cases, and preparing initial requirements ready for team review.

3. Prototyping and experimentation

It has become possible to translate a product concept description into a visual prototype or an early experience, then iterate and test it before entering a full development cycle.

4. Data and feedback analysis

AI can assist in synthesizing customer feedback, categorizing recurring issues, and summarizing vast amounts of data, helping the team spot points that warrant validation.

What is notable about these use cases is that they do not make the product manager less important; rather, they redistribute their time. Instead of spending a significant portion preparing test assets, they can dedicate more time to understanding the problem, evaluating options, and assessing outcomes.

Here, a crucial shift emerges: It is no longer necessary for a product manager to execute every technical detail themselves to test an idea, but they do need sufficient knowledge to guide the tools and evaluate their outputs.

When AI gives a convincing answer... how do we know it's right?

This is one of the points most deserving of attention, especially with the growing reliance on AI for research and decision-making.

In the episode, Amin Fadhel shared a compelling thought about the need for people who know how to question and verify AI outputs, rather than individuals who simply accept them because they sound plausible.

This was not just theoretical. The guests discussed cases where AI might provide a citation for a claim, only for a check of the source to reveal that the information isn’t there at all.

This is an issue that might seem minor in casual conversation, but it becomes far more perilous when it shapes a product decision, executive presentation, or study relied upon by the team.

A product manager might base a conclusion on an inaccurate figure, recommend building a feature based on unverified research, or share competitor intel unsupported by the primary source.

Therefore, asking the tool for sources is not enough; one must actually go back and read what they say.

AI outputs can be managed through a set of practical steps:

• Requesting original sources rather than settling for summaries.

• Fact-checking figures and citations in context.

• Reviewing the underlying assumptions behind the response.

• Distinguishing between verified facts and inferences or forecasts.

• Refraining from turning raw outputs into decisions prior to validation.

AI may shorten the time to access information, but it does not remove the responsibility to verify it.

Will demand for product managers decline due to AI?

With every technological wave, the same question resurfaces: Will increased automation reduce the need for employees?

In the episode, the conversation touched on the (Jevons Paradox), an economic principle suggesting that increased efficiency in using a resource can sometimes lead to increased consumption of it, rather than a decrease.

The idea was framed in the context of AI in an intriguing way: if an employee can complete more tasks in the same timeframe, the scope of work a team can undertake may expand, giving rise to opportunities and initiatives that were previously unfeasible to pursue.

However, this does not mean jobs will be untouched, or that every task will remain in demand in the same form. Certain roles may evolve, and specific skills may lose some relevance, while new needs arise.

For product management, a product manager could oversee a broader spectrum of experiments, working more closely with design, engineering, and marketing teams, backed by a more holistic understanding of the product.

The critical question here is not merely how many jobs will remain, but how the nature of the work within them will transform.

Skills rising in importance in the AI era

If executing tasks becomes faster, what will make one individual better equipped to leverage AI than another?

In the segment on future skills, an insightful discussion unfolded around critical thinking, logical analysis, and understanding people, behavior, and contexts.

This point warrants reflection; while AI can produce countless answers, it requires a human to know which questions are worth asking and which answers deserve trust.

Key skills that are growing in importance include:

Critical thinking: The ability to challenge outputs and avoid treating them as established facts.

Logical analysis: Deconstructing problems, understanding causes and effects, and distinguishing between what seems plausible and what is supported by evidence.

User empathy: Interpreting people's needs and behaviors, and tuning into the pain points they may not articulate directly.

Contextual awareness: Recognizing that a solution suited for one customer or market may not work for another, and that decisions are shaped by environment, circumstances, and goals.

Product sense: The ability to separate genuine problems from noise, and discerning what makes an experience valuable and distinctive for the user.

From this comes one of the finest paradoxes in the discussion: the more capable AI becomes, the more we need to understand human beings.

Technology can help us generate more options, but it cannot decide on its own which experience will make a user feel it is worth their time and trust.

From building products to embedding AI within products

This shift is not limited to product teams using AI for research, design, and development. It extends to the products themselves, as intelligent tools become part of the user experience and how work gets done.

In e-commerce, for example, the role of AI is no longer confined to writing product descriptions or suggesting ad copy. It can be integrated with store data to analyze performance, monitor operations, and streamline tasks that once required navigating across multiple pages and tools.

Salla is moving in this direction through a suite of capabilities serving merchants directly.

Chief among them is Mushamer by Salla, which enables merchants to turn text descriptions into storefront design components, with options to preview, edit, and reuse them. It also provides insights to help understand store performance and uncover opportunities to enhance the sales experience and grow the business—with feature availability varying based on product development. Discover Mushamer.

There is also the Salla AI Connector, which enables linking the store to AI assistants like ChatGPT and Claude via the MCP protocol. Once integrated, merchants can query details on orders, inventory, and sales, and even execute actions such as updating stock levels within the permissions granted to the integration. Learn about connecting with ChatGPT and Claude.

What matters here is not simply the existence of new tools, but the transition of AI from a technical experiment into the merchant's everyday workflow. This opens up a different way of thinking about product value: How do we bring technology closer to user needs while reducing intermediate steps and specialized expertise?

What will make the difference in building products going forward?

After all these shifts, it might seem that AI will make building products accessible to everyone, and that technical execution will no longer be the primary barrier it once was.

Yet ease of building does not equate to ease of success.

Companies still need to uncover genuine problems, understand customers, offer differentiated value, reach the market, and craft an experience that compels people to choose their products and remain loyal.

Concluding the discussion, two critical thoughts emerged regarding a company’s ability to sustain its competitive edge: continuous innovation, and dedication to customer experience. Even when tech tools level out and solutions become easier to replicate, a company's capacity to discover new opportunities and elevate user experience remains vital to its competitiveness.

This brings us back to the question we began with: What happens when building products becomes easier?

Perhaps the competitive edge will not belong to the one with the tool that churns out the most prototypes, or the team that executes fastest. Rather, it will belong to the team that knows which problem is worth solving, how to test it, and what makes the solution truly valuable.

In an era where building a product has become easier than ever, the hardest skill of all may be knowing what is worth building in the first place.

Discussion source: The episode "AI in Product Management" from the Zero to One podcast, featuring Amin Fadhel, VP of Product at Salla, and Abdulrahman Hareth, hosted by Ibrahim Al-Zakari.