
ThinkTank is an AI-powered Agentic Discourse Engine that bridges 19th-century colonial literature with 21st-century socio-economic data to generate historically-grounded policy analysis. The problem is clear: 2.19 million Filipinos work abroad, contractualization continues to affect millions of workers, and policymakers debate these crises without historical context. Yet Jose Rizal documented the same structures of exploitation in 1887 — and no tool existed to connect his words to today's data. ThinkTank solves this. Built on AMD's Instinct MI300X GPU using Qwen3.5-27B-FP8 served through vLLM with ROCm 7.2.0, ThinkTank orchestrates two specialized CrewAI agents: a Scholar Agent that retrieves relevant passages from Rizal's novels through a ChromaDB RAG pipeline, and an Economist Agent that searches for live labor data through web search. Together they analyze a user-submitted modern topic — contractualization, OFW displacement, brain drain — through the lens of Rizal's colonial critique. The output is a structured report featuring a historical analysis, a modern economic analysis, and a Sukdulan verdict — a culturally-framed AI judgment on a five-level scale drawn from the Filipino concept of "sukdulan," meaning the breaking point. A final policy essay written in Rizal's introspective voice is addressed directly to policymakers and educators. ThinkTank serves educators generating historically-grounded lesson modules, NGOs producing rapid policy briefs, government labor agencies contextualizing reform, and researchers conducting comparative analysis. The platform is corpus-agnostic — future versions expand to Rabindranath Tagore for India, Chinua Achebe for Nigeria, and Gabriel García Márquez for Colombia, making ThinkTank a global instrument for nations seeking to understand their present through their literary past. Deployed and tested on AMD Developer Cloud within a $100 credit budget, spending under $12 total.
10 May 2026