The TCS approach Using the exis�ng enterprise IT compute, network, and storage systems as a founda�on, the TCS AI architecture adds a second layer of founda�onal LLMs, data lakes and external data stores. Purposive and contextual AI task agents sit in the third layer along with guardrails to observe, learn and adapt. The final layer adds intelligent orchestra�on of task agents into AI-augmented work systems that partner with human employees. (See the figure below.) Employees Customers AI for AI-augmented Intelligent orchestra�on of business and IT work system purposive task agents, human-in-the-loop smart agents Purposive and contextual task agents Create and tune models, setup guardrails, observe, learn and adapt AI for AI Founda�onal LLM, data warehouse, data lakes Structured data stores, unstructured data, external data stores, analy�cs and insights Enterprise AI infrastructure (IT and AI) Core enterprise Compute, network, data Exis�ng enterprise IT and OT storage, digital products and IT and OT services pla�orms, systems of record Mul�-�ered architecture of future AI-mature businesses An effec�ve and efficient AI solu�on design is an art, with many rapidly evolving design choices and tradeoffs. TCS AI for Business Study Key Findings Report 49