Kateryna Bondar

Russia is fielding AI-enabled autonomous drones at the tactical edge despite lacking a frontier AI programme of its own, according to a CSIS report by Kateryna Bondar. Drawing on Russian policy documents, statements from senior officials, more than 150 Telegram channels, and interviews with Ukrainian operators and outside experts on the Russian armed forces, the paper traces how Moscow has stitched together a national ecosystem for unmanned systems and applied AI — from a presidential decree setting 2030 targets for compute, drone output and AI graduates, to private drone schools that fold frontline feedback into curricula within weeks. The report argues that systems such as the V2U loitering drone, which intercepted units show carry onboard compute but no command link, demonstrate functional autonomy in GPS-denied environments. It documents how Russian developers build on open-weight models from Llama, Mistral, Qwen and DeepSeek, and finds that roughly 69 percent of memory and 57 percent of processors recovered from Russian drones come from US firms. Useful for analysts tracking sanctions, export controls, and the diffusion of battlefield autonomy.

Russia is abandoning the ambition of a single, joint-style command-and-control architecture and instead fielding narrow tactical software shaped by the war in Ukraine, according to a new CSIS report by Kateryna Bondar. The paper traces the long-running ACCS programme — the Russian counterpart to the U.S. JADC2 concept — and argues that institutional inertia, Soviet-era research bodies and weak data practices have stalled the integrated vision, even as standardisation around Astra Linux and domestic Elbrus and Baikal processors continues. What is advancing, Bondar finds, is tactical: tools such as the Svod situational-awareness complex and the Glaz/Groza reconnaissance-strike workflow now compress detection-to-fire cycles, with unmanned systems running up to 80 percent of fire missions. AI sits in a support role, mature for computer vision and sensor fusion but still nascent for language tasks, and leans heavily on adapted open-weight models including Mistral, Qwen, LLaMA and YOLO. Useful for analysts tracking Russian military software, drone-artillery integration, and how wartime pressure reshapes C2 doctrine.

Ukraine's Delta platform has, in practice, delivered much of what the U.S. Department of Defense has been chasing for years under the banner of Combined Joint All-Domain Command and Control. This CSIS report by Kateryna Bondar traces how a volunteer-built digital map, started by the Aerorozvidka group in 2016 and transferred to the Ministry of Defense in 2023, grew into a cloud-hosted ecosystem now used across the Ukrainian armed forces, national guard, border service and police. The author walks through Delta's main applications: the Delta Monitor map, the Mission Control scheduler for drone sorties, the UA DroneID friend-or-foe layer, the Delta Tube and Vezha video tools, and the AI-driven Avengers system, which the ministry credits with detecting roughly 12,000 pieces of enemy equipment a week from drone and camera feeds. Bondar also covers secure messaging, mobile device management, planned analytics including a Ukrainian-language chatbot, and interoperability work with NATO systems and Link 16. Useful for defence-tech analysts, military planners and AI practitioners studying battle-management software.