brytesika-AI/1PhD-Thesis---AI-in-strategic-Decision-making

This repository contains the AI-driven Strategic Resilience Framework (AI-SRF), my PhD artifact built using the Google Agent Development Kit. It is a multi-agent governance tool designed to help South African organizations safely deploy AI and maintain digital resilience despite severe infrastructure and regulatory constraints. # AI_SRF Strategic Decision Operating System AI_SRF is a Cloudflare-native, multi-agent strategic decision operating system for board-level and executive decision support. It is designed for high-risk organizational decisions where infrastructure volatility, regulatory obligations, operational constraints, and strategic uncertainty must be reasoned about through auditable tools rather than free-form agent text. The system combines tool-based agents, strategic framework selection and blending, simulation, a digital twin, shared organizational memory, procedural learning, and a strategic narrative generator. Its primary production runtime is Cloudflare Workers, Cloudflare D1, Cloudflare KV, and Cloudflare Pages. ## 1. Project Overview AI_SRF helps an organization move from case-level reasoning to organization-level intelligence. A governed decision run can: - Retrieve organization-scoped memory before agents reason. - Select the best strategic frameworks for the case. - Run framework tools such as SWOT, PESTLE, Porter's Five Forces, Value Chain Analysis, and Scenario Planning. - Blend framework outputs into prioritized risks, opportunities, constraints, tradeoffs, and recommended strategy. - Update and use a digital twin of organizational and environmental conditions. - Simulate decisions before execution. - Enforce policy, consensus, and human approval gates. - Store memory and learning after each run. - Generate an executive narrative for boardroom communication. Core capabilities: - Tool-based agents with per-agent tool allowlists. - Automatic framework selection and hybrid framework blending. - Simulation engine with best, worst, and realistic scenarios. - Digital twin state for environment, operations, risk, and decision feedback. - Organization-scoped memory and learning logs. - Safe JSON enforcement and schema validation for tool outputs. - Narrative generator using Pyramid Principle and Situation-Complication-Resolution. - D1-backed audit replay and eve …