OBJECTIVES: To explore the value of ‘soft intelligence’, leveraged using natural language processing (NLP) techniques, as a useful source of evidence and analysis. As a case study, we deployed an artificial intelligence (AI) platform to identify and analyse global tweets on pharmacological treatments for anxiety and depression. METHODS: A search strategy comprising a list of currently available pharmacological interventions for anxiety and depression was developed and used to search for relevant tweets via Twitter’s advanced search application programming interface. Terms in the strategy included both the generic and trade names for medicines across eight different classes of drugs. We used a specialist AI and NLP platform to analyse the tweet frequency, sentiment, and key topics of discussion for qualitative analysis. RESULTS: We identified and collected 233,470 tweets, globally, over a 12-week period (27 July to 19 October 2020). The classes of drugs with the highest volume of tweets were benzodiazepines (n = 115,849) followed by selective serotonin reuptake inhibitors (n = 63,514). The lowest volume of tweets (n = 2097) was generated by drugs not belonging to a particular class (e.g., mirtazapine). The average sentiment score across all tweets was 49%, suggesting overall negative sentiment. Varying levels of sentiment were observed between different classes of drugs, most being either negative or neutral. Various topics of discussion with both positive and negative underlying sentiment were identified, coded, and summarised. Key emerging themes included people reporting specific adverse reactions, withdrawal experiences, and reasons for taking their medication. CONCLUSIONS: The findings suggest that there is potential value in ‘soft intelligence’, like Twitter, as a useful source of evidence, particularly where more robust evidence is lacking. Further, the use of AI techniques to leverage this type of data may offer an efficient means of extracting key insights from the ‘public voice’ concerning pharmacological treatments.