A unified view of where Sovereign AI investments are happening — who is funding, buying, building, and partnering — powered by a continuously updated global news pipeline.
Sovereign AI Radar turns announcements into a structured, comparable, and searchable stream of investment signals. It helps leaders track momentum by country, region, vendor ecosystem, and spend — and spot emerging patterns early.
Sovereign AI is moving from ambition to execution: national programs, cloud capacity buildouts, strategic procurements, and public-private partnerships are accelerating globally. But signals are fragmented across outlets and formats.
Sovereign AI Radar provides:
- Signal over noise: filters generic AI chatter and isolates sovereign/government-backed investments.
- Comparable data: normalizes currencies and geography so trends can be measured consistently.
- Confidence in insights: reduces duplicates and “echo coverage” to avoid inflated narratives.
- Executive visibility: a live dashboard built for quick scanning and deeper drill-down.
Each validated item is standardized into a consistent record (example fields):
- Country / Region (aligned to UN M49)
- Buyer / Program (where available)
- Vendors / Partners (where available)
- Investment value (USD) when disclosed or inferable
- Category of activity (e.g., infrastructure, model dev, applications)
- Source link + timestamp for traceability
The dashboard and dataset make it easy to answer questions like:
- Where is spend concentrating—by region, country, and category?
- Which vendors are repeatedly present in sovereign deals?
- What’s new this week — and what’s re-reported news?
A Looker Studio dashboard summarizes the latest validated signals with:
- KPI cards: total spend (USD), news volume, countries active
- Geographic views: distribution by country/region
- A searchable “Latest Announcements” table with direct links for fast verification
Sovereign AI Radar operates as a simple end-to-end loop:
- Monitor: scans Google Alerts (Atom) and Google News (RSS) feeds for relevant coverage
- Extract: uses Google Gemini to convert articles into structured “news/program” fields
- Standardize: uses a Python (Pandas) enrichment layer to normalize:
- currencies → USD
- country names → canonical formats
- country → UN M49 region/sub-region
- De-duplicate: reduces repeated coverage across outlets so metrics reflect unique events
- Visualize: updates Looker Studio for real-time exploration
graph LR
A["🌐 Data Sourcing"] --> B["🤖 AI Analysis"]
B --> C["💎 Data Enrichment"]
C --> D["🛡️ Quality Control"]
D --> E["📊 Executive Radar"]
- Traceable: every record links back to a primary source.
- Comparable: monetary values are normalized to USD; geography is standardized.
- Conservative: duplicates are removed to prevent inflated counts and misleading narratives.
- Repeatable: the process runs on a schedule with consistent rules and audit-friendly outputs.
The system produces two main datasets:
- Daily Updates: newly detected items and extracted fields (raw structured signals)
- Daily Enriched: cleaned + standardized + de-duplicated records used for analysis and dashboarding
- Google Apps Script for ingestion + daily automation
- Google Gemini (
gemini-flash-latest) for classification and structured extraction - Python + Pandas for enrichment and quality control (currency, geography, normalization)
- RapidFuzz for fuzzy duplicate detection across reworded reporting
- Looker Studio for executive visualization
- ChatGPT (
5.2 model) for AI-assisted coding
If you want to run this pipeline privately:
- Google Sheet + script: initialize the tracker and schedule daily pulls
- API key: configure the AI extraction key via script properties
- Enrichment step: run the enrichment script/notebook to standardize data
- Dashboard: connect Looker Studio to the enriched dataset
Notes:
- The system is designed to control cost via daily item limits.
- Currency conversions depend on FX availability and the chosen conversion date policy.
Potential expansions:
- Vendor & partner graph (repeat players, ecosystems, alliances)
- Program maturity scoring (intent → committed → contracted → delivered → scaled)
- Asset mapping (compute clusters, data centers, sovereign clouds, model hubs)
- Alerting (notify when new high-value signals appear in key regions)
- Open an Issue: https://github.com/buildcampmode/sovereign-ai-radar/issues
- Or comment on the LinkedIn/X post where this was shared
Built and maintained as part of the Sovereign AI Radar initiative by Saurav Chowdhary