The Scholarly Kitchen: When AI shifts from writing tool to workflow infrastructure

Hong Zhou’s February 2026 analysis in The Scholarly Kitchen examines eight major technology trend reports — from Bain, CB Insights, Deloitte, Gartner, IBM, McKinsey, Kantar, and SaM — through the lens of scholarly publishing and learning. The consistent finding across all eight is a shift in how organizations are positioning AI: not as a tool added to existing processes, but as infrastructure that requires those processes to change.

The distinction matters in practice. Organizations that treat AI as a faster writing instrument — assigning it tasks within workflows designed for human workers — typically see incremental gains. Organizations that redesign their workflows to accommodate AI as a structural participant see qualitatively different results. The reports agree that gains appear when AI embeds into end-to-end processes alongside organizational restructuring, not when it runs in parallel to a workflow that continues unchanged.

This applies directly to writing and editorial work. A team that uses AI to speed up the first draft of every piece while leaving editing, fact-checking, approval, and publishing processes untouched is automating one step in a workflow built for a different pace. A team that redesigns how stories are commissioned, drafted, reviewed, and distributed to account for AI-assisted throughput at each stage is doing something structurally different.

Several data points from the reports stand out. Deloitte found 38% of organizations piloting AI agents, but only 11% had moved to production — a gap that suggests the structural redesign required for production use is a real obstacle, not a minor configuration step. IBM’s research found 89% of consumers want disclosure when interacting with AI, which has direct implications for any editorial team producing AI-assisted content for public audiences. And 93% of executives reported that AI sovereignty — decisions about where models run and where data resides — affected their 2026 strategy, bringing it into the domain of editorial leadership, not just IT.

The article also frames trust as a design problem rather than a communications problem. IBM’s disclosure finding suggests that trust in AI systems must be built into interfaces and workflows from the start. Security and transparency mechanisms that are added to existing systems after deployment tend to be less effective than those that are foundational.

For writers, editors, and content strategists, the most directly actionable argument is the process redesign point. The question is not which AI tool to add to a workflow but whether the workflow itself should change when AI is a capable and reliable participant in it. For teams that have adopted AI tools without that structural rethinking, the gap between what the tools can do and what the organization is actually getting out of them may be larger than it appears.