Preparing for Future Product Development

Original

Most current content related to artificial intelligence focuses on how to use AI to improve products and services. This is certainly a fascinating topic-we hear more inspiring cases every day, and an increasing number of people are starting to view generative AI as a platform shift. Building a leading enterprise requires not only strategy and mechanisms but also a vital organization. However, the key focus of this article is on how product development methodologies and current practices are already evolving, and what preparations we can make. Innovations are emerging in both startups and large corporations. Dozens of new technologies, tools, and hundreds of new enterprises are born every week. Keeping up with these innovations is extremely challenging. As per the famous Amara's Law about disruptive technologies: "We tend to overestimate the short-term impact of a technology and underestimate its long-term impact."

Innovations in Product Development Methods

If we now turn our attention to how products are developed, I would argue that the internet has had the most profound impact on product development to date. Before the internet, product teams were larger, development cycles were longer, feedback collection took more time, and iterations were fewer. The internet enabled new models for product exploration, delivery, distribution, and feedback loops. We are at the center of an evolving ecosystem where front-line teams are experimenting with transforming how we develop and deliver products. Methodologies like agile, customer development, and lean startup emerged from the experiences of these early teams. This leads to a critical question: How will the rise of generative AI affect our product development approaches?

Transformations in Product Development

In some ways, generative AI represents a new and powerful tool. While this sounds simple, its practical implementation is complex. As Benedict Evans pointed out: "There’s an old saying that when we get a new tool, we first try to make it fit our old ways of working, and then later change our ways of working to fit the tool." Let’s discuss how these technologies might reshape our workflows, focusing on tools for both product development and operations.

1. Product Teams and Team Topology

Companies’ products are developed by various product teams. Since most products involve multiple teams handling different domains, "team topology" refers to how these teams are organized-a difficult yet crucial decision for current product leaders. One major constraint is "cognitive load"-the amount of information engineers can retain. Consider that many products today have tens of millions of lines of code; this is a very real challenge. While we have methods to abstract complexity, it remains a significant factor.

With new AI tools, engineers’ cognitive load is likely to decrease-perhaps substantially. This means teams could take on broader responsibilities, potentially expanding their scope significantly. Fewer and smaller product teams would bring tangible benefits to companies, including enhanced empowerment, autonomy, communication, coordination, collaboration, and job satisfaction. One obvious possibility is that companies may reduce the number of product personnel required, a trend we are already witnessing. Historically, however, as companies pursue more products and new ventures emerge, the industry as a whole experiences growth over time.

2. Team Autonomy and Initiative

Autonomy is a key factor in product team members’ job satisfaction, referring to how many resources they can control versus how many they must obtain from others. In most large organizations, relying on numerous other teams to complete meaningful work can be frustrating and disempowering. Limited autonomy can make even minor feature implementations overly complex, leading many teams to avoid tackling the most impactful problems to sidestep the extra effort of communication, coordination, and dependency management. A major benefit of next-generation AI tools is the potential for higher team autonomy, especially for engineers.

Product teams typically handle work of varying scales: from small features to medium projects and large initiatives affecting multiple teams. Many companies struggle with fostering initiative, which is crucial for addressing the most meaningful problems with high impact potential. The promise of new technologies is that autonomy will increase to a level where individual teams can not only fully own and deliver routine features and projects but also act with greater independence.

As AI reduces engineers’ cognitive load, enhances designers’ experience design capabilities, and frees product managers from administrative burdens, product teams are poised to take on more end-to-end experience responsibilities with significantly fewer dependencies.

3. Team Dynamics

Product teams are fundamentally trust-based social structures that must be cross-functional, with collaboration as the key to true innovation. Even before generative AI, many teams struggled to build and maintain trust, especially in remote settings. The question now is whether new generative AI tools will help build trust or inadvertently exacerbate issues by making it easier for people to retreat into isolation, interacting primarily with AI agents rather than cross-functional colleagues.

Healthy collaboration requires some necessary friction-the ability to express disagreements and resolve conflicting goals and constraints. Many people avoid friction, but this can stifle innovation. While tools may help cultivate trust and collaboration, there is a genuine concern that they could harm teamwork if people use them to avoid human interaction.

4. Thinking and Judgment

This area presents both the greatest potential and risk. Next-generation AI tools have the power to enhance our thinking and reasoning more than any previous tools. However, they also risk replacing essential human thinking and judgment, which could severely harm product teams and society at large. Product teams solve complex problems in ways that delight customers while benefiting the business, requiring cross-functional teams to deliberate on trade-offs. Above all, this process demands critical thinking and judgment-indeed, training product managers, designers, and engineers to navigate these challenges is one of the most important skills.

If people abandon real thinking and judgment in favor of relying on technology, we are heading in a dangerous direction-a risk that cannot be overstated. Conversely, if we ensure these tools are used to enhance the quality of human thinking, they will make a huge positive contribution.

5. Quality

Quality assurance is another domain with immense potential and risk. How do we ensure our products consistently deliver value to customers? On one hand, this has long been a challenge for many teams, requiring significant investment in test automation and ongoing maintenance of fragile automated tests. Next-generation AI-based testing tools promise to revolutionize how we ensure product reliability.

On the other hand, our current understanding of quality is rooted in deterministic products-where given inputs produce predictable, consistent outputs. For many new generative AI-based products, however, outputs are probabilistic rather than deterministic. We can no longer expect the same inputs to yield identical results. While this is acceptable in many cases, safety-critical scenarios will require new approaches to ensure proper behavior.

Ultimately, we can make educated predictions but must acknowledge our cognitive limitations. Always remember the first principle of product development: "Know what you don’t know." The future is unknowable, but we can prepare by learning emerging technologies and continuously exploring how to leverage them for our customers. Those who can quickly adapt, empathize deeply with users, and possess the judgment to tackle complex problems have always thrived amid technological disruptions-and this will hold true for the challenges ahead.