Generative AI systems increasingly mediate information discovery and recommendations. Unlike conventional search, where visibility can be approximated through ranked results, generative systems synthesize information and may produce materially different answers to similar questions. This study measures the consistency of AI recommendations across seven generative AI assistants using 84 Portuguese consumer-service brands across seven sectors. The research executed 105 prompts three times across seven AI engines with active web search, producing 2,205 responses and 588 brand-engine observations. A brand was classified as recommended when it was explicitly presented as a suitable option in at least two of three runs across at least three distinct prompts, including at least one bottom-of-funnel or persona-and-location prompt. A citation was recorded when the response included a link or identifiable reference to the brand's website or an independent source supporting the recommendation in at least two responses. The findings show substantial cross-engine fragmentation. 54.8% of brands were recommended by zero or one engine, while only two brands were recommended by all seven. The average brand was recommended by 1.77 engines. Across brand-engine observations, 37.8% included a mention, 25.3% a recommendation and 15.1% a citation. Of the 149 recommendations, 74 (49.7%) were accompanied by a qualifying citation. Measuring one engine, or measuring presence alone, will therefore misstate a brand's position. The dataset is released as a baseline against which later waves can be compared. Competing interests: this research was conducted by Destaque.ai, a commercial company that sells services in the field of Generative Engine Optimization, and therefore has a commercial interest in the phenomenon measured. No client of Destaque.ai is included in the sample, and no external party funded, reviewed or approved the study. See Section 14 of the paper.