Building a Low-Cost, Low-Code AI News Tracker for Africa’s Emerging Tech Landscape
The following is a guest post from Jordan Levinson, a participant in the June cohort of Applied AI (XD131).
I recently completed exchange.design’s Applied AI (XD131) course and can’t say enough good things about it.
I’m a digital health strategist with 10+ years’ experience partnering with governments, healthtech teams, and global organizations to design and scale digital systems across Africa, Asia, and Latin America — from food aid distribution platforms in conflict zones to mobile tools enabling millions of frontline health workers to deliver quality care in remote communities. Like many in my field, I was unexpectedly sidelined by the USAID funding shutdown, and I decided to use the time to upskill in AI applications that could bring immediate value to current and future clients.
For my final project, I set out to track and analyze AI-related news coverage in Africa — building an automated, AI-driven workflow to visualize the pace and distribution of AI’s growth across the continent. The journey is described below, and links to the final products are at the bottom of this post - check them out and let me know what you think!
Building the Automated News Bot
Browse.ai scraped targeted sources — OECD’s The AI Wonk and TechCabal’s AI coverage.
For The AI Wonk, I integrated Browse.ai with Google Sheets and used “GPT for Sheets” to summarize scraped URLs. This surfaced an important limitation that would have been very easy to miss: the tool hallucinated shockingly good summaries from just titles/snippets, without full-text access — a good reminder to dig deeper than surface outputs.
To resolve this issue for TechCabal, I used ChatGPT to walk me through writing a Google Apps Script to extract and clean full article text before summarization. This worked flawlessly — even stripping ad text — and was something I could never have coded from scratch on my own.
Screenshot of AI in LMICs news sources in Google Sheets
AI-Powered Data Analysis & Visualization
I used GPT for Sheets to tag articles by country, clean the data, and prepare it for mapping. GPT then pointed me to Flourish.Studio, where I created an interactive heatmap showing which countries had the most AI news mentions, and guided me through design tweaks to make it more engaging. I wasn’t previously familiar with Flourish, but might make it my go-to for lightweight visualizations as it was quick and intuitive and even the free tier posts interactive outputs on an easy-to-share website.
The workflow is fully automated up to the visualization stage — and could easily be adapted for real-time market tracking, policy monitoring, or competitive intelligence. The total build took about 8–10 hours, with minimal coding experience required. Given what I learned the first time, I expect it would take <4 hours to repeat.
Screenshot of final project visualization in Flourish.Studio
Final tech stack and costs:
Browse.ai for automated news scraping: Free
Google Sheets + GPT integration for processing and tagging: $20/mo paid GPT account + $29 in OpenAI API credits powering GPT for Sheets and Docs
Custom Google Apps Script for full-text extraction: Free
Flourish.Studio for interactive visualization: Free
Key strategic takeaways:
Keep focused on ‘problem first’ to avoid platform fatigue. Fancy features are abundant, but the real value comes from aligning tools to your goals, not the other way around
Low-cost is possible, but free might not be good enough. There are significant tradeoffs in data safety on free plans, and compute limits make things difficult for even one-off, small-seeming professional use cases
AI amplifies, but does not replace human skills - and common sense is still ours alone. With limited coding experience, this project would have taken me weeks to build without ChatGPT - the key was combining domain expertise with AI capabilities, not replacing either one. And as always, human oversight and an extremely critical eye are crucial.
📊 See the outputs: [News bot spreadsheet] | [Interactive map]
I would love to do more projects like this. If you work in digital health, global development, or emerging markets and want to explore how AI can streamline data collection, cleaning, and visualization— let’s talk!
