How we use AI.
We believe in being clear about what AI does in our product and how we use it responsibly.
This page explains how Narvia's AI scoring model works, what data it processes, what we store, and the steps we take to ensure transparency, accuracy, and responsible use.
What Narvia's AI does
Narvia uses artificial intelligence to analyse publicly available website content and produce an AI visibility score across eight proprietary dimensions. The scoring model identifies how clearly a company's brand, positioning, and product descriptions are structured for AI systems to parse, cite, and recommend.
Specific AI-powered features include:
- AI Discoverability scoring across 8 weighted dimensions
- Page-level analysis identifying the weakest pages and their primary issues
- Estimated impact calculations showing which fixes improve visibility most
- Connector page content restructuring integration descriptions for AI citation
- Competitor benchmarking against market averages
These features are designed to assist marketing and go-to-market teams in understanding and improving AI visibility. They are not a performance certification and should not replace professional judgement.
What we analyse, and what we do not
Narvia processes only publicly available website content submitted by the user. Specifically:
- The text content of pages submitted for analysis via URL
- Publicly visible headings, body copy, product descriptions, and navigation text
- Page structure signals such as heading hierarchy and content organisation
We do not process, access, or request:
- Internal company data, CRM records, or proprietary information
- Personal data of website visitors or customers
- Authentication-protected or gated content
- Any data beyond what is publicly accessible on the submitted URLs
Which AI systems power Narvia
Narvia's scoring engine is powered by Claude, Anthropic's AI model, accessed via secure API. Content submitted for analysis is sent to Anthropic's API for processing under Narvia's proprietary scoring prompts.
- No customer data is used to train public AI models
- Narvia's scoring prompts and methodology are proprietary and are not shared
- We do not sell, license, or expose submitted content to other customers or third parties
- API calls to Anthropic are governed by Anthropic's enterprise data processing terms
Narvia does not allow AI-generated outputs to take automated action on your website or systems. All outputs are delivered as recommendations for human review and implementation.
What we store, and for how long
Free scan: the company name, URLs submitted, overall score, and dimension scores are stored to improve Narvia's scoring model over time. The scraped page content is not stored beyond the analysis session. No personal contact data is collected or stored during a free scan.
Enterprise plans: full reports including scores, findings, and action plans are stored securely to enable monthly rescoring, delta tracking, and improvement measurement. Clients may request deletion of their data at any time.
- Scan results are stored to improve scoring accuracy across the market
- Stored data is never used for public model retraining without explicit consent
- Data is processed under EU GDPR with legitimate interest as the lawful basis
- You may request deletion by contacting lilian@narvia.io
What AI scoring can and cannot tell you
Narvia's scores are estimates of AI Discoverability readiness under Narvia's proprietary model. They are not:
- A guarantee of how any specific AI system will respond to queries about your brand
- A certification of brand health or performance
- A prediction of revenue or pipeline impact
- A universal benchmark applicable across all AI systems and contexts
AI scoring outputs may occasionally miss context not present in publicly visible page content, apply scoring criteria differently across very different industry types, or produce variance if the same URL is analysed at different times. A score of 100 means fully optimised under Narvia's model, not a universal benchmark. You remain responsible for reviewing and acting on AI-generated recommendations before implementing them.
Our commitments
Narvia is committed to responsible AI development and transparent operation. This means:
- Scoring methodology is applied consistently across all submitted sites
- We do not inflate scores to make results more commercially attractive
- We do not make claims about AI Discoverability that go beyond what our analysis can support
- Outputs are reviewed for accuracy before product updates that affect scoring
- Compliance with EU GDPR and emerging AI governance standards
We encourage users to treat Narvia scores as an informed starting point for improvement, not as a definitive measure of brand performance.
How we update this page
We will update this page as Narvia's AI capabilities evolve. The effective date at the top of the page will reflect the date of the most recent update. Enterprise clients will be notified of material changes to data processing practices.
Questions about AI, data, or compliance
Questions about this disclosure, data processing, or compliance? Email lilian@narvia.io.