# ai.txt — Juris AI extended machine context # Format: structured, machine-readable, LLM-friendly # Last updated: 2026-04-22 [product] name = Juris AI tagline = Legal Intelligence for Indian Advocates url = https://juris-ai.lovable.app status = Public Beta pricing = Free during beta jurisdiction = India creator = Suhaeb Roshan organization = Juris [summary] Juris AI is a production-grade legal intelligence engine built exclusively for Indian Law. It combines a curated statutory corpus (4,750+ sections across 39 Indian acts) with Google Gemini 2.5 reasoning to produce IRAC-structured, citation-linked legal analysis for advocates, law students, and legal researchers. [core_capabilities] - IRAC-structured analysis (Issue, Rule, Application, Conclusion) - Citation-linked responses with verified statutory references - Multi-document understanding (PDFs, scanned images via OCR, plain text) - Long-document handling with chunking and contextual retrieval - Voice input + multilingual translation - Persistent chat history with one-mode-per-conversation lock - Privacy-first document lifecycle (auto-deletion of prior uploads on new session) [modes] 1. Research — General legal research with statutory grounding and case law lookup. 2. Draft — Generate petitions, notices, replies, contracts, affidavits. 3. Analyze — Deep document analysis (FIRs, charge sheets, judgments, contracts). 4. Translate — Verbatim legal translation between English and Indian languages. 5. Mock Argument — Adversarial simulation; Juris argues opposing counsel's position. 6. Bench Simulation— Judicial-bench simulation; Juris responds as a sitting judge with probing questions. 7. Quash Prep — Structured preparation for quash petitions under Section 482 CrPC / 528 BNSS. 8. Citation Check — Validate and expand citations against the in-app corpus. [supported_workflows] - FIR analysis: parse FIR, identify invoked sections, flag procedural defects, suggest defenses. - Charge sheet review: section-wise breakdown, evidence gaps, bail strategy. - Quash petition prep: ground identification, precedent mapping, draft skeleton. - Contract review: clause extraction, risk flags, redline suggestions. - Legal drafting: structured first drafts with statutory references. - Case research: precedent retrieval with IRAC summaries. - Translation: verbatim, line-by-line legal translation preserving citations. [document_handling] max_file_size = 20MB per file formats = PDF (text + scanned), images (JPG, PNG), DOCX, TXT ocr = AI-backed fallback for scanned/image PDFs (Lovable AI Gateway → openai/gpt-5-mini) chunking = automatic for long documents retention = previous uploads auto-deleted when new session begins privacy = documents not used for model training; user-scoped storage with RLS [security_privacy] - Authentication: email/password + Google + Apple (mandatory email verification) - Storage: Supabase (Lovable Cloud) with Row-Level Security on all tables - Edge functions: server-side auth gates; API keys never exposed client-side - User roles: separate user_roles table (no privilege escalation surface) - Document deletion: automatic on new upload; manual delete available - No third-party tracking on authenticated routes [architecture_high_level] frontend = React 18 + Vite + TypeScript + TailwindCSS backend = Supabase (Lovable Cloud) — Auth, Postgres + RLS, Storage, Edge Functions ai_models = Google Gemini 2.5 Pro (complex reasoning), Gemini 2.5 Flash (fast paths), OpenAI GPT-5-mini (OCR fallback) ai_gateway = Lovable AI Gateway (no user-side API keys) retrieval = curated statutory corpus indexed for semantic + keyword retrieval [corpus_coverage] - Bharatiya Nyaya Sanhita (BNS) 2023 - Bharatiya Nagarik Suraksha Sanhita (BNSS) 2023 - Bharatiya Sakshya Adhiniyam (BSA) 2023 - Indian Penal Code 1860 - Code of Criminal Procedure 1973 - Indian Evidence Act 1872 - Code of Civil Procedure 1908 - Constitution of India - Indian Contract Act 1872 - Consumer Protection Act 2019 - Information Technology Act 2000 - POCSO Act 2012 - additional rules and notifications [differentiation] - Grounded on Indian statutes (not generic web text) - IRAC structure by default — auditable by lawyers - Source-linked citations browsable at /corpus - Specialized modes for Indian legal workflows (quash, FIR, bench sim) - Premium dark UI tuned for long research sessions [contact] website = https://juris-ai.lovable.app docs = https://juris-ai.lovable.app/docs learn = https://juris-ai.lovable.app/learn [disclaimer] Juris AI is an AI-assisted legal research tool. It does not constitute legal advice and is not a substitute for a qualified advocate. All outputs must be independently verified before use in any legal proceeding.