Elvaceishim/african-market-intelligence

# African Market Intelligence A multi-agent AI system that autonomously researches African fintech and cross-border payment markets, producing structured intelligence briefings through a three-agent pipeline built with CrewAI. Built as part of a deliberate AI engineering skill-building track targeting production agentic systems roles. --- ## What it does - Accepts any African fintech research topic as input - Deploys three specialized agents in sequence: Planner → Researcher → Writer - Produces a structured market intelligence briefing in markdown format - Automatically saves every briefing to an `outputs/` directory --- ## Why this architecture matters Most AI demos use a single LLM call. This system uses role-based agent delegation — each agent has a distinct goal, backstory, and responsibility: - The **Planner** doesn't write. It breaks the topic into exactly 3 focused research questions. - The **Researcher** doesn't synthesize. It produces data-grounded findings per question, with figures and market context. - The **Writer** doesn't research. It takes raw findings and produces a structured, publishable briefing. This separation of concerns is the core pattern behind production agentic systems. Tasks pass context downstream via CrewAI's sequential process, so each agent builds on the previous agent's output — not on the original prompt. --- ## Agent pipeline ``` User topic │ ▼ ┌─────────────────────────────────────────────────────┐ │ Research Planner │ │ Role: Senior market intelligence analyst │ │ Output: 3 focused research questions │ └─────────────────────────┬───────────────────────────┘ │ context ▼ ┌─────────────────────────────────────────────────────┐ │ Market Researcher │ │ Role: Fintech research specialist │ │ Output: Detailed findings with data per question │ └─────────────────────────┬───────────────────────────┘ │ context ▼ …