Proof of Concept— please read before relying on this data
This tool is an early prototype, not a polished product. It uses Google's public search index (via Vertex AI grounded search) to find mentions of Barbados and runs an LLM-based analysis pipeline over them. Everything you see below is auto-generated from a single weekly scan.
Known limitations
- Social-platform blind spots. Facebook, Instagram and TikTok content is poorly indexed by Google for non-logged-in users — volume from those platforms will stay low until we integrate a paid social- listening API (Brandwatch, Talkwalker, Sprinklr) or accept the ToS risk of cookie- based scraping.
- Sample, not census. Each scan returns up to ~10 results per platform × region × search term. It surfaces representative content, not every mention.
- Limited region coverage. Currently scans 12 markets. A production version should add Asia-Pacific markets, more European countries, and second-language coverage (Spanish, Portuguese, German, French in-language search).
- LLM-attributed sentiment. Sentiment, theme, and country are inferred by Gemini from short snippets. Quality is good but not audit-grade — manual review is needed before publishing findings.
- No historical backfill. Data builds over time from when scanning started. A production version would backfill from archive sources to give immediate context.
What a fuller build would add
- Paid social-listening integration for real Meta/TikTok/X volume.
- Topic taxonomy + custom monitoring queries beyond just “Barbados”.
- Competitor benchmarking against other Caribbean destinations.
- Alerting (Slack/email) when negative sentiment spikes.
- Human-in-the-loop tagging to correct mis-classified mentions.
- Multi-user collaboration with comments and saved views.
- Export to PowerPoint/PDF for client-ready briefings.