TMTom Mathewinridingbluewaves.hashnode.dev·Jul 5 · 4 min readEngineering SAGE: an anticipatory intelligence system for India's crude oil supply chainIndia imports most of its crude oil, and a large share of it passes through a single stretch of water: the Strait of Hormuz. A serious disruption there is not a market inconvenience, it is a national 00
TMTom Mathewinridingbluewaves.hashnode.dev·Jul 5 · 5 min readEpisode 4: from_pretrained, not fit: loading a worldview instead of simulating oneA live intelligence system has a cold-start problem. Before the first signal arrives, SAGE already needs to know that Jamnagar is a refinery, that a large share of India's crude transits Hormuz, that 00
TMTom Mathewinridingbluewaves.hashnode.dev·Jul 5 · 5 min readEpisode 3: Answering a crisis in 50ms by computing a future that hasn't happenedWhen a crude corridor like the Strait of Hormuz is about to close, the useful question is not "what happened" but "what happens next, day by day, to India's refinery feedstocks and strategic reserves.00
TMTom Mathewinridingbluewaves.hashnode.dev·Jul 5 · 4 min readEpisode 2: Why SAGE keeps two knowledge graphs, not oneSAGE already has a knowledge graph. Every entity, every typed relationship, every risk score lives in a Graphiti graph backed by FalkorDB. So when we added a second graph, a set of Obsidian-style Mark00
TMTom Mathewinridingbluewaves.hashnode.dev·Jul 5 · 5 min readEpisode 1: Why raw signals never touch SAGE's vector storeSAGE watches four always-on feeds for signs that India's crude oil imports are about to be disrupted: vessel movements (AIS), news and GDELT, sanctions lists, and oil prices. The obvious way to build 00
