What AI Product Managers Actually Do — and How to Hire One

AI PM demand nearly doubled in 2021 alone. But most job descriptions and hiring processes still confuse the role with a data scientist or a generalist PM. Here's what the scope actually covers — and what to look for when you hire.

Mayur Domadiya · June 12, 2026 · 6 min read

Sixty-one percent of high-performing organizations increased their AI adoption during the COVID-19 pandemic. That acceleration created a specific hiring problem: companies needed people who could translate between executive strategy, customer research, engineering teams, and data science — all at once, on AI initiatives that had no established playbook. Demand for AI product managers nearly doubled in 2021 alone, and it has kept growing. Adrian Gonzalez — an AI and ML specialist who has held multiple PM and consulting roles and lectures at MIT on AI project management, ethics, and big data — breaks down what this role actually involves and what to optimize for when filling it.

What Makes AI PM Different From General Product Management

The early days of AI in product generated a surge in demand for data scientists. As AI in product has matured, the data tasks required have expanded dramatically — data pipelining, exploratory analysis, MLOps, model deployment, monitoring — and have required new specialized roles: data engineer, data analyst, ML engineer. The AI product manager is what holds this ensemble together. Gonzalez's framing is direct: AI projects are a team sport, and the AI PM is the head coach.

The scope of the AI PM role merges three strands that most PM roles handle separately:

Strategic: Before a company commits budget to an AI initiative, someone has to assess the realistic return on investment — factoring in what AI and ML technologies can actually do, the effort to implement them, available infrastructure, and staff resources. The AI PM who can do this analysis credibly is an invaluable asset regardless of company size.

Tactical: At smaller AI-focused companies, the PM vs. product owner split may compress. The AI PM may work at sprint level: planning iterations, conducting user research, running retrospectives, demoing to stakeholders. At larger organizations, this tactical work is handled by others — but the AI PM still needs to communicate clearly across both executive and team levels.

Technical: AI PMs need a working technical foundation. The ability to discuss model trade-offs, experimentation design, infrastructure choices, and technology stack decisions is a core requirement — not a nice-to-have. Gonzalez recommends augmenting this through online certifications (Udacity, MIT), continuous reading of current AI literature, and project-based experience over academic credentials alone.

What Determines the Scope of the Role

The actual shape of an AI PM role varies significantly with context. In a small company in the early stages of AI adoption, the AI PM may need to perform data analysis when no data analyst exists, or contribute to architecture design if they have an engineering background. At a more mature organization with full ML infrastructure, the role is more focused on strategic direction and cross-functional coordination. This means the right hire depends heavily on the organizational context — not just on abstract qualifications.

Gonzalez also flags a specific risk: an unbalanced team structure makes the AI PM's job unsustainable. If an organization in the early stages of its AI journey has hired only one or two data scientists without ML engineers or AI architects, the PM will be filling gaps that should be filled by specialists. When evaluating a role or assessing a new hire's context, the composition of the existing team is as important as the PM's individual skill set. Asking about the choices already made around talent acquisition and division of labor is essential due diligence.

Portfolio Over Credentials: The Chicken-and-Egg Problem

The hiring challenge for AI PM roles is structural: most postings require prior AI experience, but the path to gaining that experience typically runs through AI roles. For candidates, Gonzalez recommends bridging this gap through academic AI projects, hackathon implementations, and cloud platforms (AWS, GCP, Azure) to develop and test AI functionalities independently. Industry-specific experience is particularly valuable — a PM who has worked in healthcare, finance, or supply chain and now wants to work in AI can leverage domain expertise to work on AI applications in those sectors. That combination of domain knowledge and AI PM skills is harder to hire for than technical credentials alone.

For hiring companies: treat portfolio and project experience as the primary signal. A candidate who has shipped a small AI implementation — even in an academic or hackathon context — has more relevant evidence than one with an AI certification and no shipped work. The certifications validate foundational knowledge; the portfolio validates execution.

How to Hire: What to Look For, What to Avoid

There is no industry standard for AI PM interviews. Gonzalez describes the range as wide: hypothetical scenario questions, hypertechnical sessions with data scientists and engineers, traditional PM-oriented discussions, and HR interviews with AI buzzword checklists. For hiring managers, this inconsistency is actually useful information — the technical interview reveals how candidates reason about model tradeoffs under pressure; the PM discussion reveals how they think about discovery, validation, and customer feedback loops.

Gonzalez's hiring recommendation is explicit: when building an early AI team and given the opportunity to hire, source people with the exact experience needed rather than betting on development. Hiring best-in-class talent is more expensive upfront and saves money overall by avoiding the mistakes that cost more to fix later. For hyper-specialized roles — ML engineers with specific architecture experience, for example — contractors can have lower total cost than permanent employees. Once the initiative is more mature and structured, it becomes more practical to develop less-experienced professionals.

Helping the executive team hire the right talent is critical to the role of an AI product manager. An unbalanced team could make your job as AI PM very difficult — pay attention to who has already been hired before accepting the role.

One consistent point Gonzalez makes about hiring is the undervalued importance of soft skills. Technical companies over-index on engineering credentials and under-invest in the communication, collaboration, and emotional intelligence that actually determines whether AI initiatives ship and whether teams stay functional under uncertainty. AI product development involves high levels of ambiguity — models behave unexpectedly, timelines shift, stakeholder expectations need constant management. The AI PM who cannot manage that interpersonal layer creates compounding problems that no technical skill fixes.

Connecting the Dots: What Good AI PM Work Looks Like

Gonzalez describes the final layer of AI PM work as "connecting the dots" — and it's the highest-leverage part of the job. When the team structure is right, the technical resources are in place, and executive support exists, the AI PM's job is to align AI initiatives with data management strategy, infrastructure evolution, and other product lines that can feed on or benefit from AI outputs.

This alignment work is where the strategic value compounds. An AI feature built in isolation may succeed as a feature; an AI feature built as part of a coordinated data and product strategy can create leverage across multiple product lines simultaneously. The AI product engineering work that actually creates that leverage requires this coordination layer — and the AI PM who can execute it is the hire that makes the difference between AI as an experiment and AI as a competitive advantage.

What This Means

The AI PM role is neither a data science role nor a generalist PM role. It's a specific hybrid that requires strategic, tactical, and technical competence — with the relative weight of each dimension determined by the company's AI maturity and team composition. For founders hiring an AI PM: start with an honest audit of what the team already has and what it actually needs. For PMs considering AI roles: the entry path runs through portfolio projects and domain expertise, not credentials alone. General AI knowledge is becoming a commodity; the premium is on the specific combination of domain depth, product judgment, and technical credibility that makes an AI initiative actually ship.