# The Impact of AI Recommendation Pages on the Software Market
The rise of artificial intelligence has transformed the way companies make purchasing decisions, particularly in software. According to a recent report by Trellner Research, three sites have produced over 215,000 pages of software recommendations specifically for AI. These pages, while extensive, raise crucial questions about the reliability and integrity of the recommendations they provide.
The Genesis of AI Recommendations
In September 2026, a study was conducted to evaluate the recommendations provided by two web-grounded AI models, Perplexity/sonar and Perplexity/sonar-pro. The goal was simple: identify the best software solutions in 380 categories, ranging from CRM to museum collection management. The results revealed some 7,534 citations from 2,055 distinct domains.
Reliability and Ranking of Sources
A key point of the study is that 59.8% of the cited sources are outside the top 100,000 most visited sites, according to the Tranco ranking. Furthermore, 23.4% are not even in the top million. This raises questions about the legitimacy and relevance of these recommendations. Indeed, a significant portion of the recommendations comes from relatively new sites, some existing only since December 2023. These sites are designed to be read by models rather than by humans, which can skew the reliability of the information provided.
The Influence of Recommendations on the Market
The fact that three sites have created over 215,000 recommendation pages suggests a deliberate strategy to influence AI models. Tech companies need to be aware of this dynamic when relying on automatically generated recommendations. A lack of discernment could lead to decisions based on biased or unverified data.
Implications for Decision Makers
For tech decision-makers, it's crucial to go beyond mere algorithmic recommendations. Source verification and critical evaluation of recommendations remain essential. Companies must ensure that the products they choose truly meet their specific needs, not just an AI-generated profile.
Conclusion
While AI recommendations can offer efficiency gains, it is vital to remain aware of potential biases and the quality of sources. Ultimately, a combination of human intuition and algorithmic advice remains the best approach for software purchasing decisions.
Let's discuss your project in 15 minutes.