What’s The Difference Between A Data & AI Product Manager & A Digital Technical Product Manager?

We agree that AI changes the technology landscape, but for some reason, extending that to say it will change the product manager's role is controversial. If the product landscape changes, doesn’t that mean the people who manage the product’s value creation will, too?

In a previous article, I wrote a Data and AI Product Manager’s job description that looks nothing like the traditional Technical Product Manager’s. It’s my most liked and viewed post to date.

I teach a Data and AI Product Management certification course that’s been taken by over 2800 students. Many go on to land AI PM roles or lead a data and AI PM organization. However, the curriculum and capabilities I teach are very different from traditional technical product management.

The problems begin when companies press digital-focused Technical Product Managers (TPM) to take over data and AI-focused products. The assumption is that even though the AI PM role differs, traditional product management capabilities will generalize, and TPMs will succeed. But that’s not the case.

Demand for these roles is rising. LinkedIn currently has twice as many job openings for Data and AI Product Managers than for Data Engineers or Data Scientists. Median salaries are higher, and the number of qualified candidates is much lower. We must acknowledge that the roles are fundamentally different, so we stop setting TPMs up for failure and upskill people from TPM or technical IC roles with the right capabilities to succeed.

Those who are prepared for an AI PM role have a high-end career path in front of them. Unfortunately, those who get set up to fail are typically laid off in less than 18 months. If you’re interested in advancing, my certification courses will set you up for success.

This is a long read, so check out the end for a TL;DR if you only need the big picture.

The 50K Foot View & Foundational Differences

We need a new breed of product manager to bridge the gap between data, analytics, machine learning, AI, business strategy, operations, customer needs, and a completely different type of product. TPMs understand software development and digital user experience. Data and AI products are a different beast.

TPMs focus on making sure things work as they’re coded. AI PMs must grapple with the inherent uncertainty of AI, where a product's reliability and ability to adapt are key.

Users interact with AI differently. They're not just clicking buttons and entering text; they're engaging in a conversation and collaborating with the AI to achieve a goal. That requires a whole new way of thinking about design.

Monetization is different. With traditional products, it's all about features and benefits. However, with AI, value is measured in outcomes. The AI PM's job is to connect a complex series of dots and estimate opportunity size upfront.

Traditional PMs follow a linear development path, even if we call it Agile. AI development is iterative by nature, responding to new data and constantly refining the model. That challenges product managers who are used to predictable timelines.

AI PMs need a unique blend of skills to navigate this new world. A technical foundation is essential. They need an understanding of AI, data science, and data engineering, but not at the practitioner’s level. They must understand what’s feasible and how each piece gets built. AI PMs must see practical limitations and know where the barriers and pitfalls are.

Still, the role has always been more than just technical chops. They must be strategic thinkers who can connect AI initiatives to big-picture business goals.

AI PMs must explain complex concepts and new paradigms in a way everyone understands. They’re the bridge between the data and AI teams and the rest of the business. That means being able to speak both languages.

As someone who’s been building and advising businesses about data and AI for over a decade, I can't stress enough how critical the AI PM role is. They’re the ones who can turn AI's potential into real-world results, and that's something businesses can’t afford to ignore.

Core Focus

The fundamental difference between TPMs and AI PMs lies in their core focus. TPMs working with digital products operate in a world of certainty. Software behaves in a predictable, deterministic manner, executing commands as programmed and delivering expected outputs. The TPM's primary goal is ensuring the product functions precisely as coded, with usability being a key focus.

In contrast, AI PMs work with inherently probabilistic products. The output of AI systems is not always completely predictable due to the nature of machine learning and complex algorithms. This introduces an element of uncertainty that a TPM doesn't deal with.

An AI PM's core focus revolves around managing this uncertainty and ensuring the reliability of the AI product. Reliability means understanding that the system will not always be 100% accurate but can still deliver value by operating within acceptable parameters.

AI PMs must consider factors like model accuracy, explainability, and potential biases to build trust with users accustomed to digital products' predictable nature. For instance, an AI PM working on a medical diagnosis tool needs to focus not only on the accuracy of the model's predictions but also on the ability to explain those predictions to doctors and patients. This ensures the AI tool is used appropriately and doesn't lead to misinformed decisions.

AI PMs excel at understanding the nuances of AI's probabilistic nature and translating that into a product that delivers value despite the inherent uncertainties. They must educate stakeholders and users on the differences between deterministic digital systems and stochastic AI systems, setting realistic expectations and highlighting the unique benefits AI brings to the table.

Most new capability requirements are extensions of this core focus.

User Interaction Paradigm

The shift from traditional digital products to AI-driven solutions changes how users interact with technology. TPMs focus on designing user interfaces that are intuitive and easy to use. Users interact with these digital products as tools, providing specific commands and receiving predictable responses. Think of a simple calculator app - you input numbers and operations, and it gives you the result. The user experience is clear and predictable.

AI PMs, on the other hand, must design for a more collaborative interaction between humans and machines. Users aren’t just giving commands; they’re engaging in a back-and-forth exchange, providing context and feedback to shape the AI's output. This shift requires the AI PM to think carefully about facilitating this collaborative relationship and building trust in a system that might not always behave in a completely predictable way.

A key challenge is managing user expectations. The probabilistic nature of AI might throw off users accustomed to digital products. For example, a user might expect a chatbot to always give the same response to a question, but with AI, the answer might vary slightly based on the context or the model’s understanding. The AI PM needs to prepare users for this and design the product in a way that makes these nuances clear.

Key areas where AI PMs need to focus to support this new interaction paradigm:

Transparency. AI PMs must provide users with insights into how the AI works, why it makes certain decisions, and what data it's using. This transparency helps build trust and allows users to understand the AI's limitations.

Explainability. It’s not enough for AI to give a correct answer. It must explain its reasoning. Users must understand the basis for the AI's recommendations, or they won’t act on them. Digital products gain trust through repeatability and predictability. AI products gain trust through explainability and transparency.

Control. Users should feel in control of the interaction and be able to intervene if necessary. This might involve providing ways to edit the AI's output or set specific parameters for its behavior.

Monetization and Value Creation

When it comes to monetizing AI products, we're entering a new frontier. TPMs are used to a world where value is directly tied to features and functionality. You build a core product, add features, and customers pay for those features. It's a pretty straightforward equation where code generates value.

However, with AI, the value proposition isn’t as obvious. Code creates very little value. Curated data is the primary value-creating asset. The internal expertise and supporting platforms that optimize model training and inference serving have significant value.

AI products create value by augmenting human capabilities, automating complex tasks, and uncovering hidden insights from data. It’s not always easy to put a price tag on those benefits because the value realization is based on an outcome vs. a task being done or work product being delivered.

Traditional monetization models like subscriptions or licensing fees might not be the best fit for AI solutions. AI PMs need to get creative and explore new pricing models that reflect the unique value AI delivers. Salesforce recently introduced a pricing model based on conversations for its AI customer service agents. This aligns the cost with the value delivered - each conversation that leads to a successful outcome generates revenue.

Key areas where AI PMs need to focus when it comes to monetization and value creation:

Quantifying ROI. This is essential for justifying AI investments and demonstrating their impact on the bottom line. AI PMs need frameworks for measuring the value of AI solutions, even when the benefits are not immediately apparent. They can achieve this by connecting AI initiatives to KPIs the business cares about: increased revenue, reduced costs, or improved customer satisfaction.

Articulating the Value Proposition. AI PMs must be able to communicate the value of AI to stakeholders in a way that resonates. They need to translate technical jargon into business language and focus on the outcomes that matter most to the organization. A clear and compelling value proposition is essential for securing buy-in for AI initiatives and ensuring their success.

Developing Innovative Pricing Strategies. As mentioned earlier, traditional pricing models might not be suitable for AI products. AI PMs need to explore new and creative pricing strategies that align with the value delivered by AI. This might involve charging per outcome, usage-based pricing, or even value-sharing models where the customer and the AI provider split the gains from the AI solution.

Remember, data is a novel asset that can be monetized multiple times over extended periods. AI PMs need to understand the unique properties of data and develop strategies to maximize its value. A successful AI product delivers novel functionality to unlock new revenue streams. AI PMs position themselves as key drivers of business growth with capabilities and knowledge that TPMs’ experience doesn’t teach.

Product Development Lifecycle

The way we build AI products is fundamentally different from traditional software development. TPMs working with digital products are accustomed to a linear development lifecycle. You gather requirements, design the product, write the code, test it, and release it. Each phase is distinct, and the process flows smoothly from one stage to the next. Even if you break the Waterfall model and do these steps in short iterations, the fundamental process is unchanged.

But with AI, it's all about iteration and experimentation. Building an AI product is like running a scientific experiment; you start with a hypothesis, test it with data, analyze the results, and refine your approach. This cyclical process continues throughout the product's lifecycle as models constantly learn and adapt to new information.

Here's where the challenges and opportunities arise for AI PMs:

Managing Expectations. AI development takes time. Training an AI model, validating its performance, and ensuring its reliability is complex. AI PMs must manage stakeholder expectations and communicate that AI products are not built overnight. Setting realistic timelines and highlighting the iterative nature of AI development is essential.

Delivering Incrementally. To offset longer delivery timelines and still meet the business’s expectations of short-term value creation, AI PMs must break initiatives down so value is delivered every 6-8 weeks. The product roadmap aligns smaller initiatives with the larger AI product and platform.

Embracing Uncertainty. AI PMs must be comfortable with nothing ever being “done.” They must also be able to make decisions with incomplete information and adapt to unexpected outcomes. AI PMs need frameworks for managing research lifecycles and innovation initiatives.

Focusing on Data. 80% of AI’s value comes from data. AI PMs must understand the importance of high-quality data and align data curation with initiatives on the AI product roadmap. They are critical in helping the business see the value in shifting from managing data to managing information and curating knowledge.

Balancing Innovation with Practicality. AI PMs need to strike a balance between pushing the boundaries of AI innovation and delivering practical solutions that meet business needs. It’s tempting to get caught up in the hype surrounding the latest AI breakthroughs, but AI PMs must stay grounded in reality and focus on building products that solve real-world problems.

AI PMs are navigating uncharted territory, requiring a unique mindset and skillset. They need to be comfortable with experimentation, embrace uncertainty, and be passionate about the transformative potential of AI. It's a demanding role, but it's also incredibly rewarding to be at the forefront of this technological revolution, shaping the future of how we interact with machines and leverage AI.

Upskilling For A Business Critical Role (TL;DR)

The AI PM role is becoming increasingly important as businesses grapple with AI's unique challenges and opportunities.

Unlike traditional software, AI products operate in a probabilistic world that requires a shift in thinking. AI PMs must manage a different type of uncertainty from their TPM counterparts.

AI PMs must prepare users who are accustomed to the predictable, deterministic nature of digital products. AI PMs must incrementally build trust in AI systems.

They must articulate a clear purpose for using AI and ensure the AI product aligns with that purpose, especially in competitive markets.

AI PMs must also guide users to collaborate with AI rather than simply dictating commands. This new interaction paradigm requires changes in product design patterns.

Data teams must be included in discovery and planning meetings to ensure the solutions address user needs and don't recreate what already exists.

Monetizing AI's value is another hurdle. AI's value often lies in augmenting human capabilities and unearthing data insights, which are more challenging to quantify than traditional software features.

AI PMs must devise innovative pricing strategies that reflect these intangible benefits. This could involve moving beyond models like subscriptions and exploring outcome-based pricing.

The development lifecycle of AI products also demands a new approach. AI's iterative, experimental nature contrasts with traditional software's linear progression. This requires a shift in mindset and a willingness to adjust to new information as it becomes available.

AI PMs must also manage expectations, emphasizing the cyclical nature of AI development, which involves ongoing experimentation and refinement. However, they must also break initiatives down so value is delivered incrementally, quarter to quarter.

AI PMs must prioritize data as the foundation of any successful AI product. They must be advocates for infrastructure investments by translating complex data requirements into impacts and KPIs that stakeholders care about. They must prove the value of data as a revenue driver and a strategic advantage for the business.