In a recent LinkedIn post, 🌀 Luca Rossi shares surprising findings from a major research project on product development processes, offering a nuanced perspective on the actual impact of AI. Rossi, who surveyed 340 teams, sat down with veteran product leader Doug Peete to discuss the results, highlighting areas where AI’s integration is either falling short or presents significant untapped potential.
The Fragility of Planning and the Rigor of Specifications
One of the primary concerns raised in Rossi’s post is the inherent fragility of product development planning. The research revealed that over 60% of teams frequently encounter missing tasks and dependencies mid-cycle. While some of this can be attributed to healthy agility, Rossi points out that a significant portion stems from inadequate initial specifications.
According to 🌀 Luca Rossi, the rigor applied to coding standards often falls short when it comes to defining product specifications. He highlights a critical gap: “only 1 in 4 engineering teams say success criteria and acceptance criteria are clear.” This lack of clarity can lead to extensive rework and delays. To address this, Doug Peete advocates for a more robust review process, suggesting the use of AI for peer-reviewing specs, alongside traditional cross-functional design reviews.
“Specs need the same rigor as code.”
The Persistent Challenge of Tribal Knowledge
Another key insight from 🌀 Luca Rossi’s analysis is the continued dominance of ‘tribal knowledge’ within product teams. The survey indicated that two out of three teams rely heavily on the knowledge held within key individuals, rather than documented information. This means critical product context, design decisions, and the rationale behind choices are often not captured in a format accessible to AI or new team members.
As 🌀 Luca Rossi notes, this reliance on individuals creates a bottleneck and hinders scalability. “Product context, design decisions, and the ‘why’ behind choices are rarely captured in a way AI or new teammates can use,” he states, underscoring the need for better knowledge management practices.
Leveraging AI Effectively in Product Management
Despite widespread individual adoption of AI tools, Rossi’s research uncovers a surprising underutilization of AI at the team level for core product management tasks. He observes that “PMs are underusing AI” for crucial functions like defining product requirements. While over 90% of individuals report using AI, fewer than 10% of teams leverage it for requirements gathering.
Doug Peete, as discussed by 🌀 Luca Rossi, exemplifies a more advanced approach. Peete utilizes AI as a continuous thinking partner, employing it for brainstorming, peer-reviewing user stories, generating interactive mockups, and streamlining the development of documentation. Rossi shares Peete’s experience of shipping a significant number of stories within a single session by effectively integrating AI into the workflow.
“Doug uses AI as a thinking partner from day one: bouncing ideas, peer-reviewing stories, generating interactive mockups, and shipping 30 stories in a single session.”
🌀 Luca Rossi also emphasizes the importance of dedicated time for ‘tool sharpening.’ While many ‘tremendously clever individuals’ discover AI workflows independently, scaling these benefits across an entire team requires intentional effort. He advocates for allocating specific time within each development cycle for teams to collaboratively build and refine shared AI workflows and documentation.
“Make time for tool sharpening — most orgs have ‘tremendously clever individuals’ who figure out AI workflows on their own, but the challenge is scaling that across the whole team.”
The insights shared by 🌀 Luca Rossi, informed by his extensive research and discussion with Doug Peete, provide a critical look at the current state of AI adoption in product development. They suggest that while AI’s potential is vast, realizing its full benefits requires a more strategic and rigorous approach to planning, specification, knowledge management, and team-wide AI integration.
📝 About This Content
This article is based on insights shared by 🌀 Luca Rossi on LinkedIn.
📅 Originally posted on May 18, 2026 | View original post on LinkedIn →