About GuardianAI
The privacy-first AI violence detection platform for K-12 schools and university campuses. We turn the cameras you already have into a real-time alerting system — without ever recognising a face.
Last updated: April 26, 2026
What is GuardianAI?
GuardianAI is school-safety software that detects physical violence on existing CCTV cameras within 1–2 seconds, using a two-stage computer-vision pipeline: YOLO11n-Pose extracts a 17-keypoint COCO skeleton per person, and a CTR-GCN spatiotemporal graph classifier scores each two-second window of motion. The classifier never sees raw pixels — by construction, it cannot identify any individual student. We sell it as an annual per-camera subscription to K-12 districts and university security teams worldwide.
Who runs GuardianAI?
GuardianAI is built and operated by Abdraman Tursynov (founder, CEO) and a small distributed team of computer-vision engineers and school-safety advisors. The product is the result of an 18-month research project at TripleAcodex (the legal entity behind GuardianAI) that asked a single question: can we cut the lag between a fight starting and an adult arriving from 8–12 minutes to under 2 seconds, without storing a single biometric record?
The answer turned out to be yes — but only by abandoning the standard playbook of facial recognition and pixel-level CNNs, and instead building a pose-only pipeline that is architecturally incapable of identifying anyone. That decision is the foundation of every other choice we've made.
What problem are we solving?
Every K-12 school in the United States already has cameras. Most don't have anyone watching them in real time. The average administrator learns about a physical altercation 8–12 minutes after it starts — through a student tip, a teacher walking in, or an injury at the nurse's office. By that point, the fight is over and the school has only post-hoc reports.
Traditional CCTV is a forensic tool: it tells you what happened after it happened. GuardianAI is the live alerting layer on top of those existing cameras. A pose-based AI model watches every stream 24/7 and pings the school resource officer (SRO), vice principal, or designated on-call adult when it sees fighting, hitting, or shoving — usually within 1–2 seconds of the first contact.
What makes us different?
| Choice | Why it matters |
|---|---|
| Pose-based, not face-based | The model operates on (x, y, confidence) skeleton coordinates only. No biometric record exists at any layer of the system. Compatible with FERPA, COPPA, GDPR Article 9, and Russian 152-FZ. |
| On-premise inference | Detection runs on a Jetson Orin Nano (or RTX appliance for larger sites) installed inside the school's network. Video never leaves the building. The school keeps the raw recording; we keep nothing. |
| Works on cameras you already own | Any RTSP-capable IP camera or NVR — Hikvision, Dahua, Axis, Avigilon, Reolink, Hanwha. Roughly 95 % of K-12 cameras installed since 2018 qualify. |
| Per-camera flat pricing | $200–$600 per camera per year, all-in. No per-incident fees, no per-alert fees, no "enterprise tier" gate. A 30-camera school pays $9,000–$15,000 a year. |
| Bilingual product (EN + RU) | The dashboard, alerts, and documentation ship in English and Russian. Schools in CIS countries get the same product as schools in the US. |
Where do we operate?
We sell to school districts and university campuses worldwide. Active deployments and pilots are concentrated in three regions:
- United States — primarily public school districts in K-12, plus several private school networks. We provide a one-page summary your attorney can compare against your existing CCTV vendor's contract.
- European Union — GDPR-compliant deployments, with a Data Processing Agreement and DPIA template available on request.
- Russia and the CIS — full Russian-language product and 152-FZ-aware data flow. The on-prem appliance keeps all video and metadata inside the school's network.
How is the company organised?
The legal entity is TripleAcodex. The product brand is GuardianAI. Sales contracts, invoicing, and DPAs are issued in TripleAcodex's name. The website, social profiles, and product UI all use GuardianAI.
How can I get in touch?
- Sales and demos: hello@guardianai.tech
- Data subject requests (GDPR / CCPA / 152-FZ): privacy@guardianai.tech — see also our data request portal
- Press, partnerships, and analyst briefings: hello@guardianai.tech
- Source overview: github.com/guardianai-tech
Where to read more
- How GuardianAI is deployed in K-12 schools — full deployment recipe, pricing, FAQ
- How GuardianAI scales to 1,000-camera campuses — clustered architecture for universities
- Technology: pose-based detection — stage one of the pipeline
- Technology: spatiotemporal graph network — stage two of the pipeline
- Pricing — per-camera tiers and pilot terms
- Privacy policy — full data handling disclosure
Russian / Русская версия: /ru/about.