01 — The Problem
Entrepreneurs and business leaders make decisions based on signals - new regulations, market shifts, competitor moves, emerging trends, unmet demands. But these signals are scattered across dozens of sources: news outlets, government gazettes, social media, industry reports, community forums.
No individual can monitor all relevant sources consistently. By the time most people notice an opportunity, the early-mover advantage is gone.
This problem is especially acute in African markets where:
- Information is fragmented across many small sources
- Data isn’t aggregated the way it is in developed markets (no Bloomberg for Zimbabwean SMEs)
- The time between signal and opportunity is often shorter (less competition, faster markets)
- Informal information networks (WhatsApp groups, word-of-mouth) contain signals that never make it to formal publications
02 — Context
Details to be added by Vincent.
03 — Why I Built It
Details to be added by Vincent.
04 — Approach
AI Opportunity Intelligence is a system that:
- Ingests information from multiple sources - news, social media, government publications, market data, job postings, tender notices
- Analyses signals using AI to identify patterns, emerging trends, and potential business opportunities
- Scores opportunities based on market size, competition, timing, and relevance to the user’s interests
- Delivers insights through a dashboard and WhatsApp notifications - surfacing what matters before it becomes obvious
The core insight: what looks like scattered information to a human brain becomes identifiable patterns when processed by AI at scale.
05 — Architecture
┌─────────────────────────────────────────────────────┐
│ Data Sources │
│ News APIs │ Social Media │ Gov Publications │ RSS │
└──────────────────────┬──────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ Ingestion Pipeline (Celery) │
│ Fetch → Clean → Deduplicate → Store │
└──────────────────────┬──────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ AI Analysis Layer (LangChain) │
│ Classify → Extract Entities → Score → Cluster │
└──────────────────────┬──────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ Opportunity Engine │
│ Pattern Detection → Scoring → Ranking → Alerts │
└──────────────────────┬──────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ Delivery Layer │
│ Dashboard (FastAPI) │ WhatsApp Notifications │
└─────────────────────────────────────────────────────┘06 — Technology
Python + FastAPI - Chose for the AI/ML ecosystem. FastAPI gives async performance without sacrificing Python’s AI library access.
PostgreSQL - Relational database for structured opportunity data, with full-text search for signal matching. Considered vector databases but structured queries were more important for this use case than semantic search.
Celery + Redis - Background task processing for ingestion pipelines. Sources are fetched on schedules without blocking the API. Redis handles task queuing and caching of recent results.
LangChain - Orchestrates the AI analysis pipeline. Handles prompt chaining, structured output parsing, and model switching. Allowed rapid iteration on the analysis prompts without rewriting pipeline code.
OpenAI API - Used for classification, entity extraction, and opportunity scoring. The reasoning capability of GPT-4 class models made it possible to identify non-obvious connections between signals.
07 — My Role
Details to be added by Vincent.
08 — Challenges
Details to be added by Vincent.
09 — What Worked
Details to be added by Vincent.
10 — What Didn’t
Details to be added by Vincent.
11 — What I Learned
Details to be added by Vincent.
12 — Current Status
Details to be added by Vincent.
13 — Demo / Repository
Details to be added by Vincent.
14 — What’s Next
Details to be added by Vincent.