QORVYNA Documentation
AI Assistant
QORVYNA Chat is a broadly useful assistant for writing, learning, planning, research, technology and safe digital life. Every request goes through a secure server-side gateway using the OpenAI-compatible endpoint, model and API key configured by the admin, with environment fallbacks when that is not available.
What it can do
- Explain difficult topics, improve writing, summarise material and build practical plans.
- Help with computers, phones, apps, AI, troubleshooting and source code.
- Read attached text and common source-code files; inspect images without exposing private image bytes to the coding model.
- Explain phishing, scams, privacy, malware and device security with calm defensive guidance.
- Reply naturally in English, বাংলা or Banglish.
Who you are talking to
The assistant is QORVYNA AI, in every mode and every language. Ask it who it is and it tells you. Ask which AI model or company's technology is behind it and it answers truthfully and specifically, from the configuration your request actually used — it does not dodge the question and it does not guess. By default that model is GPT OSS 120B, an open-weight model published by OpenAI and served by Groq; QORVYNA is the product you use, not the maker of the model.
What it won't do
- Claim it opened, ran, tested, previewed, committed or deployed something when no real tool result exists.
- Help with harmful activity: credential theft, malware creation, account takeover, DDoS, phishing infrastructure or detection evasion.
- Invent reporting websites, phone numbers, citations, commits, files or test results.
- Ask for or repeat passwords, tokens, codes, private keys, ID numbers or payment details.
How your messages are protected
Before your message reaches the model, a PII redaction layer removes emails, phone numbers, one-time codes, passwords, card numbers, addresses and ID-like data. If the request is off-topic or harmful, the model is never contacted at all.