Polygraf AI is an AI protocol that identifies if digital content is reliable, trustworthy, and/or AI-generated.

Applied to online reviews, our free Chrome browser extension, Polygraf Trust Engine, uses our AI protocol to holistically analyze review content, product features, seller & buyer behavior, and provide you with an objective AI-generated star rating and trust score so that you can shop with confidence. Sellers can rely on the fact that reviews we analyze, good or bad, are genuine feedback from the customer experience and not from malicious actors looking to damage the brand. Satisfied customers also mean minimal returns.

Polygraf Trust Engine uses real-time behavioral AI in conjunction with sentiment analysis to analyze reviews. Our algorithm is holistic and also uses other parameters when determining content reliability & trustworthiness. Our analyses are in real-time and dynamic – as users write reviews for products, our AI updates Trust scores, snapshots & ratings at the reviews, products, and seller levels. It’s also always learning, so the more reviews we analyze – the more accurate and nuanced feedback you get.

Polygraf Trust Engine is presently available for products and reviews on Amazon, BestBuy.com and Walmart.com but will soon be available for a variety of e-commerce platforms in the US and globally.

Polygraf Trust Engine generates trust scores at 2 levels: (a) At the individual review level: We calculate our trust score for each review by using our advanced AI engine. Our AI engine considers a range of factors including the history and reputation of the writer, the language used in the review, the writer's tonality, and the consistency of the review with other reviews for the same and competing products. Our AI uses these factors to generate a holistic score that indicates the level of trust you can place in that review. (b) At the product level: This is an overall trust score that is based on an aggregate of the reviews analyzed by Polygraf. This score is dynamic and changes in real time as new reviews are added.

Trust scores are calculated from 1-100%. A trust score below 45% means that a review cannot be trusted, one between 45-72% indicates potential issues with trustworthiness & malicious intent, and one above 72% means a review is reliable. A trust score measures the reliability of a review, irrespective of whether it is a positive or negative review. It is not a judgment on the quality of the product or seller.