Vefaro — Multi-Tenant AI Sales Bot SaaS, Headless Crawler & WooC by Yury ShuhayeuVefaro — Multi-Tenant AI Sales Bot SaaS, Headless Crawler & WooC by Yury Shuhayeu

Vefaro — Multi-Tenant AI Sales Bot SaaS, Headless Crawler & WooC

Portfolio project
Yury Shuhayeu
Architected, designed, and launched Vefaro (vefaro.com) — a full-scale multi-tenant SaaS platform that turns any business website, online store, or document price list into an autonomous AI sales and lead-capture assistant in under 2 minutes with zero coding or prompt writing.
Core Engineering and Technical Architecture • Headless Browser Crawler and Document Parser: Built an automated ingestion pipeline that renders JavaScript-heavy websites (SPAs, Wix, Tilda, custom stacks) in a real browser to extract services, pricing, and FAQs. Includes a drag-and-drop parser for PDF, Excel, and Word price lists to enrich the RAG knowledge base automatically. • Zero-Hallucination Pricing Engine: Engineered strict retrieval guardrails ensuring the assistant never invents numbers or discounts — every quote is strictly grounded in verified store data and cites the exact source page URL. • Pre-Configured WordPress and WooCommerce Plugin Generator: Built a dynamic packaging system that compiles a custom WordPress plugin per user with their site ID and credentials pre-embedded for zero-config activation, plus automated daily WooCommerce product catalog and price synchronization. • Frictionless Bot-First Onboarding: Visitors paste their URL and test a live, trained sandbox of their own bot with mouse-driven controls before signing up or entering a card (5-day full trial). • Multi-Tenant SaaS Billing, Agency Tier, and Telegram Ecosystem: Complete subscription infrastructure (Single, Multi-Site, and 25-bot White-Label Agency plans up to 25,000 conversations/mo), 90-day affiliate tracking, promo engine, and deep Telegram integration (@VefaroBot) for passwordless login and instant lead delivery.
Performance, Localization, and Privacy • High-Speed Stack: Next.js 16 frontend, Python (FastAPI) async backend, PostgreSQL, and DeepSeek LLM inference hosted on dedicated German infrastructure (GDPR-first, zero third-party ad trackers). • 9 Native Languages Out of the Box: Full platform and chat localization across English, German, Spanish, French, Italian, Polish, Portuguese, Turkish, and Russian. • Zero PageSpeed Impact: Single-line async embed widget delivering replies in under 2 seconds while keeping client Core Web Vitals at 95–100.

0 likes · 0 views · Nov 11, 2026

Built a multi-tenant AI SaaS platform (Next.js 16, FastAPI, PostgreSQL) featuring a headless site crawler, PDF/Excel price parser

Tags