# ZySec AI

> ZySec AI builds sovereign AI in Abu Dhabi: platforms, and the defences around them, for governments, defence and regulated industries. Your data never leaves your infrastructure, and neither does the model that learns from it. ZySec AI is an Infinia Technologies company, founded in 2024, with 50+ engineers and teams in Abu Dhabi, Hyderabad and Delaware, live in the UAE, India and Africa. Lead: Sovereign AI. Secured and governed. Sign-off: Building Trusted Intelligence.

ZySec AI's products form one sovereign AI stack. Each product is a layer, deployed on-premise, air-gapped or in sovereign cloud.

## The stack, by layer (top to bottom)

- Your people: ask, approve and decide. Consequential actions pass a human approval gate.
- Govern & prove · AgentSight (See Control Prove): AI agent identity and governance; AI governs AI, including the agents inside C3. Every agent tool call checked against an Ed25519-signed policy before it runs, blocked in-process when out of scope, with hash-chained audit and MFA for high-risk actions.
- Govern & prove · ScaleRisk (One Platform One View of Risk): AI risk and posture management; one control set mapped to the EU AI Act, ISO/IEC 42001 and NIST AI RMF, with GCC frameworks built in and evidence captured once.
- Act · Autonous (Enterprise Sovereign AI OS): digital employees with citations, no-code workflows with approval gates, tamper-evident audit.
- Act · Autonous IntOps (Enterprise Intelligent Search): the AI-First Data Lake; every archive answerable, every answer traced to file, page and passage.
- Act · C3 (Cyber Command Center): the agentic SOC; agentic SOAR and an AI security team that triage and investigate every alert above the existing SIEM, with people making the final call.
- Reason & remember · RelataDB (Native AI-Database): open-source knowledge engine with an ontology, graph, BM25 and vector search, bi-temporal agent memory and cell-level policy. https://www.relatadb.dev/
- Reason & remember · Token Factory (by Infinia Technologies): enterprise LLM inference inside the perimeter, shared by every agent and team.
- Ingest & enrich · OpenETL: brings data in from every source and prepares it for AI. To be released as open source.
- Ingest & enrich · ThreatMap: 20+ threat-intelligence sources through one normalised API, ready for AI agents over MCP. https://threatmap.co
- Your estate: existing EDR, firewall, identity, email, cloud and SIEM are kept and connected.

The reference architecture, with how the layers connect, is at https://zysec.ai/platform/

## Proof

- Running today with ADI Foundation, DDSC, Apeiro Holding, Oxinus Holdings, SAIF (an IHC platform), Coredge and ESAL Tech, and a state law-enforcement agency in India.

- An intelligence picture that took two to three days now takes minutes, with Autonous IntOps at a state law-enforcement agency in India, on the agency's own servers.
- On-premise deployment in six to eight weeks; air-gapped in eight to twelve. Zero egress: no telemetry, no callback, no internet dependency.
- ISO 27001 certified; SOC 2 and PCI DSS-ready products.

## Research and engineering

- HyperMind: Claim Resolution and Reputation Tracking for Trustworthy Multi-Agent LLM Aggregation. Accepted, CIKM 2026, Rome.
- AgentInSight: signed tool manifests, fail-closed enforcement and tamper-evident audit for AI agents. Accepted, NeurIPS 2026 Workshop AIWILD (poster).
- CyberPod: Entity Recognition and Multi-hop Criminal Network Intelligence for Investigative Analysis (2026).
- Open security models with more than 200,000 downloads. SecurityLLM (ZySec-7B) is cited in the Foundation-sec-8B technical report by Cisco Foundation AI and Yale. https://huggingface.co/ZySec-AI/SecurityLLM
- Saqr: a 27.8B-parameter multimodal cybersecurity model built by ZySec for Infinia Technologies. https://saqr.iamsaif.ai

## Recognition

- Best Emerging AI Company of the Year 2025, Indian Business Excellence Awards (ANI, July 2025).
- Showcased its sovereign, agentic AI security stack at GISEC Global 2026 (Zawya, press release).

## Links

- [Website](https://zysec.ai/): the full story, by layer
- [Research](https://zysec.ai/research/): papers, open models and datasets, citations, and the six research areas behind the products
- [Platform](https://zysec.ai/platform/): the reference architecture, how the products connect, and deployment, models, integration, security and assurance
- [Autonous brochure (PDF)](https://zysec.ai/assets/briefs/zysec-autonous-brief-2026-09.pdf)
- [C3 brochure (PDF)](https://zysec.ai/assets/briefs/zysec-c3-brief-2026-09.pdf)
- [ScaleRisk brochure (PDF)](https://zysec.ai/assets/briefs/zysec-scalerisk-brief-2026-09.pdf)
- [Open models and datasets](https://huggingface.co/ZySec-AI)
- [Full text for AI assistants](https://zysec.ai/llms-full.txt): products, engineering, research areas, publications and FAQ in one file
- Contact: hello@zysec.ai