# From needs assessment to usability testing: evaluating the AinoAid™ chatbot for domestic violence support

**Authors:** Catharina Vogt, Stefanie Giljohann, Natalie Köpsel, Sandra González Cabezas, Jarmo Houtsonen, Ainhoa Izaguirre Choperena, Anna Juusela, Emanuel Tananau Blumenschein, Margarita Vassileva

PMC · DOI: 10.1186/s12905-025-04202-3 · BMC Women's Health · 2025-12-12

## TL;DR

AinoAid™ is an AI chatbot designed to help domestic violence survivors access support services safely and anonymously, with user feedback showing it is effective but needs improvements in empathy and personalization.

## Contribution

The paper introduces and evaluates AinoAid™, a multilingual, data-secure AI chatbot for domestic violence support, validated through mixed-methods user research across multiple European countries.

## Key findings

- Users rated perceived safety as the chatbot's strongest feature.
- Participants suggested improvements in empathy, engagement, and personalized guidance.
- The chatbot was seen as a safe and accessible gateway to support services.

## Abstract

Survivors of domestic violence (DV) often encounter barriers when accessing professional support services. These barriers arise, for example, from uncertainty in identifying violence, limited knowledge of support options, and psychological barriers such as fear or shame. To address these challenges, the European project IMPROVE developed AinoAid™, a data-secure and multilingual website with an integrated chatbot. Developed in collaboration with psychotherapists, the chatbot provides low-threshold, anonymous, empathetic access to information on domestic violence support services, risk assessment and legal procedures. Ultimately, AinoAid™ seeks to serve as a safe, informative gateway to protection and support.

Employing a mixed-methods design, Study 1 involved interviews with 80 DV survivors from Austria, Germany, Finland, France and Spain to explore their openness and needs concerning an AI chatbot for 24/7 support. Furthermore, Study 2 was a German evaluation survey of 669 users, assessing perceived safety, usability, utility and interaction quality of AinoAid™.

In Study 1, interviewees predominantly reported limited prior chatbot experience, yet expressed remarkably positive attitudes toward the potential use of a chatbot tailored for DV support. They emphasized the importance of data safety, chatbot accessibility and usability, offering suggestions on chatbot content tone and promotion. The evaluation in Study 2 indicated a positive overall rating of the chatbot by users, with perceived safety emerging as the most highly rated feature. Study participants suggested improving the chatbot’s empathy, engagement, input comprehension, follow-up integration, and memory of previous statements. Additionally, users called for more personalized information, actionable guidance and direct links to relevant support services.

Our evaluation confirms that the AI chatbot AinoAid™ functions as a safe, accessible, and effective gateway to protection and support for DV survivors. Its perceived great strength, its impartiality due to its non-human nature, presents, on the other hand, also a significant challenge in meeting the high expectations for human-like, empathetic, and dynamic conversations in such a sensitive context. However, we anticipate that sustained usage and continuous training will even further enhance the AI ability to fully meet the evolving needs of domestic violence survivors.

The online version contains supplementary material available at 10.1186/s12905-025-04202-3.

## Full-text entities

- **Chemicals:** chatbot (-)
- **Species:** Homo sapiens (human, species) [taxon 9606]

## Full text

_Full body text omitted from this summary view._ Fetch the complete paper as Markdown: https://tomesphere.com/paper/PMC12817862/full.md

## Figures

3 figures with captions in the complete paper: https://tomesphere.com/paper/PMC12817862/full.md

## References

21 references — full list in the complete paper: https://tomesphere.com/paper/PMC12817862/full.md

---
Source: https://tomesphere.com/paper/PMC12817862