Objective To explore the correlation between the usage intensity and functional preference of the artificial intelligence assistant “Xiaoyibao” and anxiety symptoms and quality of life among patients with pancreatic cancer. Methods A total of 140 patients who were diagnosed with pancreatic ductal adenocarcinoma in a tertiary grade A hospital from March 2024 to September 2025 and used “Xiaoyibao” were enrolled. One-off assessments were carried out from April to May 2026. Clinical data were extracted from electronic medical records. The Hamilton anxiety scale (HAMA) was adopted to evaluate anxiety, and the European organization for research and treatment of cancer quality of life questionnaire-core 30 (EORTC QLQ-C30) was used to assess quality of life past week (T₁). The baseline EORTC QLQ-C30 score (T₀) was obtained by asking patients to recall their status within 1 week before the first anti-tumor treatment (or within 2 weeks after diagnosis for those who had not received anti-tumor treatment). Participants were divided into a high-frequency group (cumulative usage＞24 times, n＝69) and a low-frequency group (cumulative usage≤24 times, n＝71) based on the median cumulative usage times. They were also stratified into a psychological support group (n＝70) and non-psychological support group (n＝70) according to whether they used the psychological support module “Xingma Ailiao”. ANCOVA was used to compare HAMA score and EORTC_T₁ score between groups. The multiple linear regression was used to analyze the dose-effect relationship between cumulative usage frequency and HAMA score. Results ANCOVA showed that after adjusting for age, sex, tumor stage, KPS score and disease duration, there was no statistically significant difference in the adjusted mean HAMA scores between the high-frequency and low-frequency groups. After adjustment for the aforementioned covariates and the baseline EORTC_T₀, the difference in adjusted mean EORTC_T₁ scores between the high-frequency and low-frequency groups reached marginal statistical significance (P＝0.050). Compared with the non-psychological support group, psychological support group had lower adjusted mean HAMA score (P＝0.024) and EORTC_T₁ score (P＝0.002). The multiple linear regression showed a marginally significant negative dose-effect relationship between cumulative “Xiaoyibao”usage times and HAMA score (each additional 10 usages corresponded to a reduction of approximately 0.68 points in HAMA score, P＝0.050). Stratified analysis showed that this dose-effect trend was significant in the high-frequency group (Spearman ρ = −0.380, P＝0.001), but not in the low-frequency group. Conclusions The use frequency of “Xiaoyibao” exceeding a certain threshold and the use of its psychological support function help reduce anxiety levels and improve quality of life in patients with pancreatic cancer. For the promotion and functional optimization of “Xiaoyibao”, the psychological support module should be prioritized.