CCTV surveillance cameras are ubiquitous across modern institutional environments: school corridors, industrial checkpoints, military garrisons, and retail storefronts. However, in the vast majority of deployments, they serve a purely passive archival function — video feeds write continuously to on-premise hard drives, and footage is audited strictly post-incident, after physical damage, injury, or legal violations have transpired. The human operational bottleneck is structural: security monitoring staff are cognitively unable to sustain vigilant observation across dozens of concurrent feeds 24/7.
Kazakhstani startup Protector AI (LLP “ProtectorAI”) engineered a fundamentally different paradigm: overlaying artificial intelligence onto existing camera hardware without rewiring infrastructure or replacing optics. The proprietary computer vision platform performs real-time neural stream analytics, compressing threat detection latency down to 20 seconds while eliminating up to 95% of manual archive investigation overhead.
Origins: From Social Media Pain Points to Regional Pilots
The venture originated in early 2024. Three longtime colleagues — Zhasulan Salybayev (CEO), Nuras Nogayev, and Aidar Saparov (COO) — isolated a systemic institutional friction surfacing repeatedly in investigative media: instances of abuse and theft inside social care facilities for individuals with disabilities, occurring despite operating camera surveillance.
“Conventional surveillance records incidents post-factum. An operator cannot maintain focus on dozens of video streams 24 hours a day. Our platform operates as digital vision and an intelligent assistant, filtering out background noise. Crucially, executive decisions always remain with human operators: AI does not replace security teams, it makes their response instantaneous,” emphasizes Protector AI co-founder Zhasulan Salybayev.
The engineering team piloted initial algorithmic models directly within regional disability service facilities before expanding deployments into early childhood education (Altyn Besik kindergarten in Petropavlovsk) and North Kazakhstan regional schools. The geographic selection was deliberate: in 2025, the region entered the national “red zone” for juvenile delinquency metrics, and proactive visual analytics delivered a measurable reduction in escalations.
The initiative soon garnered backing from the Prime Minister of the Republic of Kazakhstan and was selected for nationwide scale by supervisory ministries. Within two years, the core engineering team expanded from the three founders to 12 specialized computer vision and software engineers.
Platform Architecture: Six Specialized Neural Vision Engines
Unlike proprietary enterprise solutions mandating wholesale camera replacement with specialized sensor arrays, Protector AI deploys over any standard RTSP-compliant infrastructure. The software architecture incorporates six specialized neural detection modules:
- Behavioral Event Analytics: Real-time detection of physical aggression, brawls, altercations, and bullying patterns during early escalation stages;
- Critical Threat Detection: Instantaneous recognition of bladed and firearm weapons, open flames, and smoke plumes;
- Restricted Zone Perimeter Security: Tracking unauthorized breach lines and loitering behaviors outside operating hours;
- PPE Safety Compliance: Automated monitoring of occupational health and safety protocols across construction and industrial facilities (hard hats, goggles, reflective vests);
- Access Control & Watchlist Screening: Biometric facial matching against authorized institutional watchlists and law enforcement databases;
- Automatic Number Plate Recognition (ANPR): Optical license plate parsing and unified facility vehicle movement tracking.
For industrial manufacturing and extractive enterprises, the engineering team launched a dedicated product line — Protector AI · Stream. This service resolves complex regulatory compliance bottlenecks under the e-ondiris system for Kazakhstan’s Registry of Domestic Manufacturers: the agent captures on-premise RTSP video streams and serves a secure, read-only web-player via encrypted WireGuard tunnels for state audit inspectors without compromising sensitive internal corporate network topologies.
Enterprise Deployments: Key Figures and Footprint
To date, Protector AI secures more than 90 enterprise and institutional clients. Landmark production deployments include:
- 55 public secondary schools across Kazakhstan, headlined by the flagship implementation at School No. 107 in Astana;
- Military installations of the National Guard of Kazakhstan: Visual algorithms detect physical aggression between personnel at the arm-raise stage, preceding impact;
- Government Ministries: Ministry of Labor and Social Protection of Kazakhstan, Ministry of Emergency Situations, and regional administrations across Akmola and North Kazakhstan;
- Telecom and Cloud Infrastructure Partners: Transtelecom JSC, QazCloud, and Iswim.
Regulatory Certification and Cybersecurity
Within Enterprise and B2G markets, verified data governance represents an absolute prerequisite. Protector AI successfully completed exhaustive state information security testing: Accreditation Certificate No. KZ.T.01.E1472, Information Security Functional Audit Protocol No. I/F-036 dated December 30, 2024, and full national certification under ST RK ISO/IEC 15408-2-2017 (“Common Criteria for Information Technology Security Evaluation”).
Industry Recognition and International Expansion
The startup achieved top honors at Central Asia’s premier technology summit, Digital Bridge 2026, and was crowned “Best AI Startup” during the Digital Qazaqstan Awards 2026 in Shymkent. The commercial model leverages recurring SaaS subscription tiers, air-gapped On-Premise deployments for secured defense installations, and custom neural vision module development.
Over the subsequent 12 months, the company’s development roadmap focuses on integration with the unified Ministry of Internal Affairs warrant database, deepening hazardous zone computer vision models for metallurgy and oil & gas, and launching collaborative pilot deployments across Uzbekistan and Kyrgyzstan alongside regional integration partners.
Explore verified startup metrics, engineering capabilities, and leadership contact details in the Protector AI profile on RAEM.KZ Startup Radar.
