GeoFaham: specialist agents beat general coding agents on disaster maps
The point
GeoFaham is a multi-agent system for geospatial disaster analytics. It splits work into specialist agents for vector operations, map feature retrieval, imagery discovery, and raster analysis, coordinated by an Orchestrator. A Response Store keeps full outputs outside the LLM context window, so agents exchange compact identifiers instead.
Why it matters
For builders, specialization and keeping artifacts out of context can beat general coding agents on cost.
Key details
- GeoFaham-Bench has 40 tasks grounded in 17 real world disaster records
- Scored 0.940 versus 0.879 for the next best baseline, with 7.5 times fewer tokens
Original ~10 min → 25 sec here Source: arXiv cs.AI, Nov 3, 2026