From Process-oriented AI to Shared Organizational Intelligence

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Generative artificial intelligence is entering organizations along two paths that are advancing at different speeds. In the first, professionals adopt tools on their own initiative to write, research, analyze documents, prepare proposals, build presentations, and support decisions. In the second, slower path, the organization tries to define how the technology should connect to processes, what information it may access, who is accountable for the outcomes, and how impact will be measured. This difference creates a paradox: a company may have many AI users and still lack an organizational AI capability. Gains remain local; prompts, criteria, and histories stay tied to individuals; results vary across functions; and leadership cannot distinguish real productivity from a simple sense of speed. Experimentation is not the problem - it is an important discovery mechanism. The limitation appears when useful experiences fail to become process, institutional knowledge, governance, memory, and learning. Based on this diagnosis, this white paper proposes Process-Oriented AI: technology takes on explicit roles inside understood workflows, with a baseline, authorized sources, human accountability, and performance indicators. Because isolated processes can still reproduce departmental silos, the more mature stage is described as Shared Organizational Intelligence: a logical and institutional layer that preserves context, authorized knowledge, rules, permissions, memory, indicators, and learning across different processes. It is not a single AI, a mandatory vendor, or a central database with unrestricted access.