Xcapit Labs
7 products in production — validated against our own client engagements
Bonum, Privacy, Shelter, ArgenTor, AiSec, OrchestAI, xNinja: the SaaS layer we built by solving real client problems first, then turning them into product. Same engineers, same standards, same ISO 27001.
OrchestAI
OrchestAI is Xcapit's enterprise platform for multi-LLM orchestration with governance. Route requests across Claude, GPT, Gemini, and Ollama with cost-optimized strategies, signed HMAC-SHA256 audit chains for tamper-evident compliance, versioned agent catalogs with promotion gates, and full on-premise deployment. Data never leaves your datacenter.
Key Features
- Multi-LLM routing (Claude, GPT, Gemini, Ollama)
- HMAC-SHA256 signed audit chain
- On-premise deployment (Docker/K8s)
Technology Stack
XNinja
XNinja is a multi-agent SaaS platform for automated penetration testing for enterprises and SMEs. LangGraph-powered AI agents orchestrate Nmap, Nuclei, and TestSSL for comprehensive discovery and vulnerability scanning, with automatic compliance mapping to ISO 27001, NIS2, BSI IT-Grundschutz, DSGVO, and TISAX frameworks.
Key Features
- Automated Penetration Testing
- 5 Compliance Frameworks
- Multi-Tenant Architecture
Technology Stack
Bonum
Bonum is a closed-loop digital wallet SaaS platform for corporate benefits. Companies distribute non-remunerative benefits digitally, employees pay via QR at affiliated local businesses, and the community benefits from a virtuous economic cycle. The platform provides full traceability, saves 80% of administrative time, and delivers 35% more effective value to employees.
Key Features
- QR payments at local businesses
- SaaS admin panel with analytics
- Multi-tenant architecture
Technology Stack
Shelter
Shelter is Xcapit's SaaS infrastructure for distributing impact assets at scale — subsidies, humanitarian aid, corporate incentives. Built on blockchain for end-to-end transparency, Shelter powers configurable disbursement rules, real-time tracking, and multi-chain settlement. Validated through the UNICEF-backed AidLink pilot in Cusco, Peru, with 100% completion rate on every targeted disbursement and near-instant settlement.
Key Features
- Configurable disbursement rules
- Multi-chain smart contracts (EVM)
- Real-time tracking & traceability
Technology Stack
Privacy
Xcapit Privacy is a platform for machine learning on data that remains encrypted at all times. Using Fully Homomorphic Encryption (CKKS scheme), organizations can collaboratively train AI models — for fraud detection, medical research, risk analysis — without ever exposing their sensitive data. Not even the platform can see it.
Key Features
- 256-bit FHE encryption
- Blockchain governance (Arbitrum)
- 15+ ML algorithms on encrypted data
Technology Stack
ArgenTor
ArgenTor is an intelligent multi-agent AI framework built entirely in Rust. ReAct reasoning, code intelligence with AST analysis for 4 languages, precise diffs, 25+ rule code review, automated TDD, and autonomous dev teams with 8 workflows. Cost-aware model routing across 14 LLM providers, A2A protocol, WASM sandbox, and ISO 27001/42001 compliance. 14 crates, 1514 tests, and 85K+ lines of code.
Key Features
- WASM-sandboxed plugin execution
- Centralized MCP protocol proxy
- Human-in-the-loop approval workflows
Technology Stack
AiSec
AiSec is an AI agent security analysis framework with 35 specialized agents that scan, detect, and auto-remediate vulnerabilities across AI systems. Featuring Docker-sandboxed execution, AI-CVSS scoring, policy-as-code governance, and SARIF/CI integration. Validated through the OpenClaw audit where it found 63 findings — 4.2x more than traditional scanners.
Key Features
- 35 specialized security agents
- Docker-sandboxed execution
- Auto-remediation engine