NekaWatch: AI Website Monitoring SaaS

Context

Classic website monitoring requires either checking pages by hand or building a parser tied to the site's HTML structure: CSS selectors, XPath, and extraction logic configured for each specific page. That setup needs a developer, and it breaks when the markup changes.

NekaWatch approaches the task differently. The user points at a page and describes in natural language the information that matters: a price, availability, a date, a status, a quantity, a deadline, or several of these at once. The platform analyzes the page itself and extracts the described values.

NekaWatch is a commercial production SaaS, fully developed and operated by NekaSoft.

What we owned

NekaSoft owned the product completely, from concept to production, and continues to develop it.

Solution

The central feature: the user does not need to understand the page's HTML at all. No hunting for CSS selectors, no XPath, no picking DOM elements, no per-field parser. You describe the data in words, and the AI analyzes the page and turns the matching values into a structured result.

• One monitor can track several described values on a page at once
• Right after creation the service shows what it found, so you confirm the right thing is tracked
• Every check produces a structured snapshot of the described data, not a raw page diff
• Snapshots are compared field by field: what changed (was → became), what appeared, what disappeared
• Checks run on a schedule; current state, found changes, and history live in the dashboard
• Retries and error handling separate a real data change from a temporary failure to fetch or analyze the page
• Notifications by email and Telegram; webhook delivery for external automated workflows
• Full SaaS around the engine: accounts, social login, plans and limits, Paddle billing

Traditional scraping is a parser tied to markup: URL, CSS/XPath configuration, maintenance on every layout change. NekaWatch is a different pipeline: URL plus a natural-language request, AI extraction into structured data, scheduled comparison, notification. Semantic extraction makes monitoring less dependent on the page's markup, so a redesign usually does not break a monitor.

Screenshots

Result

NekaSoft designed, built, and launched a production SaaS in which the entire journey is self-service. Typical scenarios: prices, stock availability, competitor data, vacancies, tenders, deadlines, and other structured information published on web pages.

• Sign up with email or via Google or GitHub
• Create a monitor: a URL plus a plain-words description, several tracked values if needed
• Set a schedule and get structured snapshots of exactly the described data
• See specific changes, field by field: was → became
• Browse current state, found changes, and history in the dashboard
• Get notifications by email or Telegram, or wire changes into external processes via webhook
• Upgrade to a paid plan and manage the subscription; billing runs on Paddle
• Use the product in English, Russian, or Serbian

Production status

NekaWatch runs in production at nekawatch.com: registration, monitors, scheduled checks, notifications, paid plans, and billing all work self-service.

The product is NekaSoft's own SaaS: the team operates it and continues to develop it.

Technology

NekaWatch runs on Next.js, Node.js, and TypeScript with Supabase. Extraction is built on the OpenAI API / LLM and turns page content into structured data; notifications go out by email, through the Telegram API, and via webhooks; subscriptions and billing run on Paddle.