AI is no longer something product teams “plug in” to a few places. It’s something they are reorganising around. AI capabilities are being embedded directly into product, CMS, and newsroom teams as part of the core workflow. As workflows change, so do the teams around them. 

The early phase of AI adoption often relied on centralised teams: small groups of specialists tasked with exploring use cases, building prototypes, and evangelising internally. That model is now breaking down.

In its place, we’re seeing a distributed approach take hold. While sometimes there are specialist AI teams within each department, most teams have AI directly integrated into them as part of the core product work evolving.

This shift reflects a simple reality: The value of AI is highest when it’s applied at the point of execution, not managed at a distance. And central teams recognise much of this has to be done at brand level, not centrally.

Alongside this, organisations are becoming much more precise about where AI actually delivers value.

Finding where AI is truly useful

One product leader told me about a project they’ve recently undertaken with their newsroom to break down workflows and see where AI, mostly within the product run CMS, could help. A consistent pattern emerged and they found up to 40% of time is spent on tasks AI can meaningfully support: formatting, enrichment, packaging, and optimisation. 

At the same time, a much smaller percentage of work remains firmly human — interviews, source development, and original reporting. By freeing up time on processes many journalists don’t love, they can do more of the work that creates real, unique stories and data.

This clarity is changing prioritisation. Instead of broad experimentation, teams are focusing on the highest-impact processes. Impact is (finally) starting to become more measurable so it’s easier to find defined sets of workflows where AI can drive immediate efficiency or quality gains.

At the same time, team structures themselves are starting to shift. Where product teams may previously have operated with larger groups of engineers supporting a single product manager, there is a move toward smaller, more focused units, often just two to three engineers per product manager.

This is partly driven by efficiency but also by the changing nature of the work. Product managers are increasingly expected to prototype ideas themselves using AI tools, reducing the need for engineering-heavy discovery phases and enabling faster iteration before development resources are committed.

Changes expected at the top

Something else that recently came up in conversation with the INMA Product & Tech Advisory Council is the shift is happening at the leadership level.

The expectation for senior product and editorial leaders is changing rapidly. It’s no longer enough to set direction and delegate execution. Leaders are now expected to engage directly with AI tools, understand their capabilities and limitations, and participate in much more granular discussions about implementation.

Conversations are moving “into the weeds” because that’s where many of the real decisions now sit. How a tool performs in practice, how it integrates into workflows, and where it breaks down are no longer details that can be abstracted away.

Taken together, these shifts point to a broader transformation: Product organisations are becoming more embedded, more technical, more operationally focused, and leaders are becoming more “hands on.”

Tools that can help

If we break product work into stages, below are some of the tools you may want to consider:

• Requirements (Claude Artifacts, ChatGPT, MCP + Jira).

• Design (Figma MCP, Figma Make).

• Code (GitHub Copilot).

• Testing (Playwright, Maestro).

• Prototyping (Lovable, Replit, Emergent).

• Orchestration (agent roles: orchestrator, QA, frontend, backend).

Across this, agents may execute parallel tasks, but humans need to supervise the checkpoints.

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

Banner photo: Adobe Stock By Pakorn.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .