A global intelligence dashboard that keeps its AI local

What it does

WorldMonitor is a TypeScript dashboard that ingests 500+ curated news feeds and 65+ external data providers—covering finance, energy, climate, aviation, cyber, and military signals—and renders them as AI-synthesized briefs, risk indices, and map layers. It offers six thematic variants (world, tech, finance, commodity, happy, energy) from a single codebase, served via web or as a native desktop app built with Tauri 2. The interface pairs a 3D globe with a flat WebGL map and supports 24 languages, including RTL layouts.

The interesting bit

The project is stubbornly local-first: you can run the entire AI pipeline through Ollama with no API keys, and the desktop app ships as a single Tauri binary that switches between variants in-app. That same local-first philosophy means missing credentials—like a flight API token—surface a credentials required message instead of synthetic placeholder data, which is a rare bit of honesty in dashboard demos.

Key highlights

• Local AI by default: Supports Ollama, Groq, and OpenRouter, but can run entirely offline via Ollama and Transformers.js in the browser.
• Dual map engine: Combines globe.gl (Three.js) and deck.gl (MapLibre GL) with 56 layer types for spatial intelligence.
• Single codebase, six skins: The full, tech, finance, commodity, happy, and energy variants are environment switches, not separate repos.
• Protocol Buffer contracts: 276 protos across 34 services with sebuf HTTP annotations, suggesting a heavy emphasis on structured API contracts.
• AGPL-3.0 with commercial escape hatch: Copyleft by default, but the author offers separate commercial licensing for teams that need proprietary use.

Caveats

• Desktop CI is selective: While the Tauri app is stable, the current release pipeline only pre-builds full and tech variants; other flavors require building from source.
• Security findings are documented: A 2026 disclosure revealed IPC command exposure and renderer-to-sidecar trust boundary issues; the project acknowledges these and publishes a security policy, but the details are worth reading before deploying sensitive data.
• Data fidelity depends on keys: Many feeds work out of the box, but features like live flight prices require specific API tokens, and the app deliberately refuses to fake the missing data.

Verdict

Analysts, journalists, and security researchers who need a self-hosted intelligence hub will find this far more capable than a typical news aggregator; if you just want a clean RSS reader or a simple stock ticker, the 56 map layers and 276 protobuf contracts are probably overkill.