Just as Japanese Kanban methods transformed manufacturing with "just-in-time" systems, companies are now using similar principles for generative AI to enhance efficiency. Sastry Durvasula of TIAA explains the firm's AI initiative, "Research Buddy, " which generates reports for Nuveen asset management using real-time data. By activating AI only when required, TIAA optimizes both timing and cost, given the high expense associated with generative AI. Forrester analyst Mike Gualtieri acknowledges the value of just-in-time AI but notes that its cost-benefit ratio varies across situations. A million-dollar AI spend, for instance, is justified if it saves significantly more. Organizations generally avoid huge costs by using pre-trained models, reducing the need for pricey data scientists. Retrieval-augmented generation (RAG) services help minimize AI costs by enhancing quality and relevance without custom training.

Vendors like Google offer ready-made RAG solutions, allowing enterprises to focus on software development rather than AI complexities. SAIC's introduction of Tenjin GPT exemplifies using AI strategically across the company. It enhances internal workflows, supports customer service, and even aids software development. CIO Nathan Rogers emphasizes the goal of democratizing AI access for timely decision-making. Max Chan of Avnet challenges the "just-in-time" label, preferring an approach that strategically applies AI where it yields meaningful results and cost efficiency. AI's computational demands dictate that it be used judiciously for business benefits. While just-in-time AI sometimes lacks a human oversight loop, TIAA mitigates this by embedding AI responsibility guidelines, ensuring analysts review AI outputs before use. Durvasula also highlights the "just-in-case" approach, where timely information is provided to professionals as needed without delays. Generative AI implementations, like the incremental changes brought by Japanese manufacturing techniques, revolutionize operations and outcomes when thoughtfully executed, whether through just-in-time, just-in-case, or strategic smart applications.