Frequently asked questions.

VritantAI is a multi-tenant agentic SaaS platform built for e-commerce brands. It combines two engines: a Shopping Agent that turns conversations into catalog-verified sales, and GEO (Generative Engine Optimisation) to improve your brand's visibility and accuracy in AI search results.

No. For Shopify, setup is a webhook configuration in your admin panel. For WooCommerce, it's a plugin install. The shopping agent is a single script tag. No engineering team required.

Shopify and WooCommerce are fully supported with native integrations. Other platforms (Magento, BigCommerce, headless builds) are available on the Pro plan via our REST API.

Yes. Every plan includes a 14-day free trial for one product. No credit card required to start.

VritantAI runs citation benchmarks across Perplexity, ChatGPT (GPT-4o), Gemini, Claude, and Microsoft Copilot. Each model is queried independently with no fallback chain in the benchmark runner, ensuring accurate per-model citation rates.

The GEO score is a weighted composite: Schema Markup accounts for 50%, Page Performance for 30%, and Entity Coverage for 20%. Scores above 80 indicate strong AI visibility. Scores below 60 typically indicate critical missing structured data.

AI models re-crawl and re-index at different cadences. You may see improvements in Perplexity within days. ChatGPT and Gemini can take 2–4 weeks for index updates to reflect. Run a new citation benchmark 4 weeks after your first batch of patches for a meaningful comparison.

Yes. Every patch goes through a human-in-the-loop review queue. You can view the full diff, approve, reject (with an optional rejection note), or edit the patch before approving. Nothing is deployed without your explicit approval.

No. The widget runs inside a Shadow DOM, which is a browser-native encapsulation boundary. Your existing CSS cannot bleed in, and the widget's styles cannot bleed out. It is safe to install on any storefront.

The hallucination guard cross-checks every product ID against your live catalog before surfacing results. If a product is out of stock or has been delisted, it will be silently stripped from the agent's response. The catalog is re-synced regularly to stay current.

Basic color and theme customisation (light/dark, brand colors) is available on all plans via the Settings → Widget page. Full CSS injection and white-labeling is available on the Pro plan.

Yes. You need a verified WhatsApp Business Account through Meta to use the official WhatsApp Business API. This is a Meta requirement, not a VritantAI one. Meta's verification process typically takes 2–5 business days.

WhatsApp Commerce is available on Growth and Pro plans. It is included in the plan price with no separate per-message fee from VritantAI. Note that Meta may charge for certain template message types.

Yes. Upgrades are immediate and prorated to the day. Downgrades take effect at the end of your current billing cycle, so you retain full access to your current plan's features until then.

Your agent widget will gracefully degrade to a 'limited availability' state. It will not crash or show errors to your customers. You'll receive an in-app notification and can upgrade your plan or wait for the next billing cycle to reset.

Yes. Convert is priced in INR and billed through Razorpay for customers in India. International customers are billed through Paddle, our Merchant of Record, which handles their local currency. You receive an invoice for every charge.

No. All data is strictly isolated per tenant using PostgreSQL Row-Level Security (RLS). Every database query is automatically scoped to your tenant ID. Your catalog, conversations, audit history, and hallucination events are never accessible to other tenants.

Data is stored in PostgreSQL via Supabase with pgvector for embeddings. Data at rest is encrypted. You can request data export or deletion at any time from your account settings.

No. Your data is never used to train third-party AI models or shared with LLM providers for training purposes. LLM API calls use your data only as query context for the specific session and are not retained by the provider for training.