AI-driven product management in IT: a fact-based perspective

What the article is about

Published in December 2025, this Medium article by Akshay Reddy builds its argument around concrete data rather than trend assertions. The central observation is that while 72% of organizations have adopted AI in at least one business function, only 18% report measurable bottom-line impact — figures drawn from McKinsey’s 2025 survey. The gap between experimentation and actual results is the problem the article sets out to address.

Key data points

Several statistics establish the scale of the challenge. Ninety-five percent of AI pilots fail to deliver measurable business impact. More than 60% of product managers lack AI capabilities. Organizations with structured AI training programs succeed at rates 28% higher than those without. External vendor partnerships succeed at twice the rate of internal builds. The article also notes that 90% of firms have employees using personal AI tools, but only 40% have officially purchased LLM subscriptions — a sign that adoption is often bottom-up and uncoordinated.

Case studies

Klarna’s AI assistant handled two-thirds of customer service interactions in the first month after launch, maintaining satisfaction scores comparable to human agents. Netflix shifted from collaborative filtering to foundation models for personalization, illustrating how infrastructure investment in AI compounds over time rather than delivering immediate returns. GitHub Copilot delivered measurable productivity improvements through a phased rollout — not wholesale adoption. Microsoft embedded governance and ethics frameworks into Copilot from the start. JPMorgan paired its IndexGPT project with workforce upskilling, treating the tool and the capability-building as inseparable.

Key framework

The article proposes a five-phase adoption roadmap: identify candidate workflows, run a limited pilot, scale what shows value, establish governance, then expand into innovation. It describes this as a product-led approach — treating internal AI initiatives the same way a PM would treat an external product, with defined users, success metrics, and iteration cycles.

The skills path is organized by time horizon: in the short term, AI literacy across the team; in the medium term, UX design for AI-assisted interactions; in the long term, the ability to coordinate agentic workflows.

Who it is useful for

Product leaders and IT executives navigating AI adoption in enterprise environments, particularly those dealing with pilot fatigue, uncoordinated experimentation, or internal resistance. The article is most useful for PMs who need to build an evidence-based case for structured AI integration rather than defending ad-hoc tool use.