xTwins is the platform we use to ship specialist agents — fast. Six are already in production for QA and Ops (Alberto, Catalina, Afonso, Matilde, plus Marco and Íris in Ops). Bring your domain, your runbooks, your APIs — we wire the next one, grounded in your project, hosted in the EU.
The first wave of AI tools sold one chatbot, hoping it could cover every job. We don’t. xTwins is a platform for building one agent per domain — grounded in that domain’s knowledge, wired to that domain’s tools, and held to that domain’s definition of done. The @register_agent mechanism lets us ship a new specialist in weeks, not quarters.
Reads a Jira Story — title, description, acceptance criteria — and drafts the user story, fills the gaps, and writes the Gherkin scenarios. Catches the edge cases your team forgets.
Turns Gherkin into runnable Ruby step definitions on Ritain’s proprietary QAT library. Reuses what already exists in your repo via RAG; only writes the steps you’re missing.
Parses any HTML in seconds and ranks Selenium XPaths by stability against UI churn. Returns alternates so your suites keep running through a redesign.
Native iOS and Android. Reads the Appium element hierarchy the way a mobile engineer reads code — picks selectors that survive an OS update.
Assists field technicians with explainable, source‑cited diagnosis. Piloted with telco operators on legacy B/OSS stacks — cutting incident resolution from days to minutes.
Classifies, prioritises and routes incoming tickets across Jira, Azure DevOps and GitLab. Multi‑label, taxonomy‑aware, with duplicate detection.
Sales research that reads your CRM. Compliance that knows your runbooks. Field service that ingests your manuals. Bring the job description — we wire the StateGraph, the prompts, and the RAG.
Three weeks to a working agent in your stack. No prompt‑engineering bootcamp, no model fine‑tuning before you see value.
A LangGraph StateGraph routes the request through the right specialists — RAG, generation, sibling agents, and back to you as a Jira comment. Every step is logged; every output is marked.
xTwins reached TRL 6‑7 in November 2025 with live pilots inside telco operators and Ritain.io’s own QA teams. The numbers below are operational, not benchmarks.
Native plugins for the issue trackers. Open protocols for the rest. Nothing leaves the EU except the OpenAI API call, under DPA with zero retention.
All services on AWS ECS Fargate, eu‑west‑3 (Paris). OpenAI traffic only via API, under DPA with zero retention. Data never leaves the EU.
Every project gets its own Qdrant collection. Context for one tenant never leaks into another. GDPR erasure removes embeddings too.
Limited‑risk classification. Every artifact marked “_generated by x‑twins_”. Inputs sanitised against prompt injection.
Built to the ISO/IEC 42001 AI management standard — governance, risk and lifecycle controls for every agent in production.
xSU = xTwins Service Unit. One xSU ≈ one agent action. Balance is managed per organisation and per project, with hard limits before the request fires.
Use the form for product conversations, pilot requests, partnerships or implementation questions. If there is a real use case, this is the right entry point.
xTwins is built by teams already working with enterprise operations, QA and AI delivery. Reach out if you want to discuss a specialist agent for your own domain, stack or project constraints.
Three weeks from brief to a working agent inside your stack. Talk to a human whenever you want.
