Ethical AI statement
Our commitment to responsible AI development and use
Last updated: October 2026
AI in our work
In our work we use AI assistants such as Claude (Anthropic) and ChatGPT (OpenAI), and AI meeting notes through Sembly. They help us find knowledge, analyse and write faster. A person reviews everything we deliver. We handle confidential information in line with our terms and privacy policy. If you have requirements about which tools we use, talk to us. We can also work in an environment of your choice.
Our approach to AI
At Narrate, we believe AI should strengthen human expertise, not replace it. Our AI tool Sada helps impact practitioners scope their work. Its recommendations are transparent, explainable and under human oversight. Sada is currently available only to invited pilot users.
Sada AI transparency
What data Sada uses
Sada analyses your responses to structured questions about your project or organisation. It uses: your text inputs, selected categories and frameworks, and anonymised patterns from similar projects (with consent).
How Sada makes recommendations
Sada uses large language models. With them, it finds relevant impact measurement frameworks and suggests suitable indicators. It also points out possible stakeholder groups and recommends ways to collect data. All recommendations are based on established practice in impact management.
Limitations of AI advice
Sada cannot: replace professional judgement, guarantee outcomes, fully understand complex organisational contexts, or make ethical decisions on your behalf. Sada’s recommendations should be reviewed by experienced practitioners.
Human oversight
Every Sada interaction puts human review first. We explain why we recommend something. Users can reject or change suggestions. A person stays involved in important decisions. And we regularly check Sada’s output for quality and fit.
Bias mitigation
We actively work to reduce bias in Sada. We train on many different impact frameworks and sectors. We regularly check recommendations for fairness. We encourage critical evaluation of the output. And we take in feedback from users in different contexts.
Data privacy in AI
Your Sada conversations are encrypted in transit and at rest. They are never used to train other companies’ models without consent. We keep them only as long as the service needs, and delete them on request. See our Privacy Policy for full details.
Ongoing evaluation
We evaluate how Sada performs. We collect user feedback, review quality and check for bias and fairness. We also follow research on AI ethics. We are committed to responsible AI and welcome feedback at impact@narrate.works.