Responsible AI
AI Governance Policy
Effective date: June 14, 2026
Purpose
Roast My Funnel uses AI to support website audits, simulate buyer personas, summarize conversion friction, and generate recommendations. This policy explains how we aim to use AI responsibly and transparently.
What the AI does
The service may analyze public website content, screenshots, visible copy, pricing cues, forms, layout, trust signals, and checkout flow. It may then generate synthetic buyer persona reactions, attention-span observations, objection summaries, scores, and prioritized fix lists.
Synthetic buyer personas
Buyer personas are simulated research tools. They are not real people, do not represent guaranteed customer behavior, and should not be treated as statistically valid market research. They are designed to expose likely objections and usability friction before real visitors experience them.
Human responsibility
AI-generated reports are decision support. Users remain responsible for reviewing recommendations, checking accuracy, considering brand, legal, privacy, accessibility, regulatory, ad-platform, and compliance context, and deciding whether to implement any change. Do not publish or rely on report language without human review.
No professional advice
Reports may mention legal, privacy, accessibility, pricing, advertising, user experience, or business concepts, but they are not legal, tax, financial, medical, employment, security, accessibility-certification, or other regulated professional advice. Consult qualified professionals before relying on output in regulated or high-risk contexts.
Known limitations
AI output may be incomplete, incorrect, outdated, biased, overly confident, or unsuitable for a specific market. The system may miss hidden checkout steps, gated pages, private customer context, offline sales dynamics, analytics data it cannot access, legal requirements, ad policy requirements, or accessibility issues that require specialist testing.
Data protection
We aim to minimize personal data in AI workflows, avoid unnecessary sensitive data processing, and use vendors subject to appropriate contractual and technical safeguards. Users should not submit confidential, private, sensitive, or unauthorized pages for auditing. Public page content may be processed by AI model providers and infrastructure vendors as needed to generate a report.
Prohibited uses
The service must not be used to attack websites, bypass access controls, scrape private systems, generate deceptive claims, discriminate against protected groups, infer sensitive traits, make automated decisions with legal or similarly significant effects, or create unlawful, harmful, defamatory, infringing, or abusive content.
Quality controls
We may use input validation, public-URL checks, prompt constraints, policy instructions, logging, report review workflows, abuse detection, rate limits, and vendor safeguards to reduce harmful or unreliable output. These controls reduce risk but cannot eliminate all errors.
Quality review and repair
Reports may pass through automated quality review and a controlled repair pass before publication to catch broken, incomplete, contradictory, unsupported, or low-specificity output. This review is designed to improve reliability, but it does not make reports professional advice or guarantee correctness.
Advertising and compliance limits
Reports may suggest changes to copy, claims, pricing, proof, disclosures, forms, and calls to action, but they do not certify compliance with FTC rules, ad-platform policies, accessibility standards, privacy laws, consumer protection laws, industry-specific regulations, or local market requirements. Users are responsible for review before publishing.
Monitoring and improvement
We may review system quality, failure cases, hallucination risks, abuse patterns, and customer feedback to improve reliability and safety. Significant AI workflow changes should be documented internally before production release.
Contact
AI governance questions can be sent to [email protected].